drkw’s technical strategistsusage increases sharply for larger firms with $70 billion and above...

Sending out clear trading signals

5002

ju

ly/aug

.technicalanalyst.co.uk wwwThe publication for trading and investment professionals

DrKW’s TECHNICAL

STRATEGISTS

Outlook for bunds Volatility arbitrage fundsFimat’s Rami Habib

discusses the strategies

Commodity currenciesWhich are the

genuine articles?

DeMark indicator signals

major reversal

July/August 2005 THE TECHNICAL ANALYST 1

© 2005 Clements Biss Economic Publications Limited. All rights reserved. Neither this publication nor any partof it may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic,mechanical, photocopying, recording or otherwise, without the prior permission of Clements Biss EconomicPublications Limited. While the publisher believes that all information contained in this publication was correctat the time of going to press, they cannot accept liability for any errors or omissions that may appear or loss suffered directly or indirectly by any reader as a result of any advertisement, editorial, photographs or othermaterial published in The Technical Analyst. No statement in this publication is to be considered as a recommendation or solicitation to buy or sell securities or to provide investment, tax or legal advice. Readersshould be aware that this publication is not intended to replace the need to obtain professional advice inrelation to any topic discussed.

CONTENTS 1 FEATURES

Sending clear signalsMax Knudsen of Dresdner Kleinwort Wasserstein

in London discusses his team's client focused research product that looks to provide

clear, user friendly trading recommendations.

Bund futures Warning of major reversal

Gordon Kolling, technical analyst atCommerzbank in Frankfurt, alerts us to a

significant DeMark signal on bund futures.

Intermarket analysis Marc Zaffran of Societe Generale in London

explains their 'Market Behaviour andHistorical Patterns' - a tool for forecasting

short-term FX rates.

JUL/AUG

>12

>15

> 39

>

> >

WELCOMEAlgorithmic execution strategies (AES) are increasingly being adopted by investment

managers and hedge funds as a means of reducing both transaction and commissioncosts. However, until now there has been little data available on the full extent of AESusage. In this issue we present the results of a Banc of America Securities sponsored

survey - the first of its kind to report on AES application across buy-side equity firmsin the US.

We also look at how Dresdner Kleinwort Wasserstein has developed a web-basedproprietary market sentiment tool to provide their clients with daily market calls and

timing recommendations across a range of asset classes. PIA, (Price InformationAdvantage), is DrKW's attempt to leverage their in-house technical analysis expertise

in the competitive prime brokerage market.

Matthew Clements, Editor


Editor: Matthew ClementsManaging Editor: Jim BissAdvertising & subscriptions:Louiza Charalambous Marketing: Vanessa GreenEvents: Adam CooleDesign & Production: Paul Simpson

The Technical Analyst is published byClements Biss Economic Publications LtdUnit 201, Panther House,38 Mount Pleasant, London WC1X 0AN

Tel: +44 (0)20 7833 1441Web: www.technicalanalyst.co.ukEmail: [email protected]

SUBSCRIPTIONS

Subscription rates (6 issues) UK: £150 per annumRest of world: £175 per annumFor information, please contact: [email protected]

ADVERTISING

For information, please contact:[email protected]

PRODUCTION

Art, design and typesetting by all-Perception Ltd.Printed by The Friary Press

ISSN(1742-8718)

INDUSTRY NEWS Special feature: Algorithmic trading survey

MARKET VIEWS Gold: Long-term trend will holdFTSE 100: First Fibonacci target at 5450Bunds: DeMark indicators warn of major reversal

TECHNIQUES DeMark Retracements and TrendlinesIntermarket analysis: Forecasting short-term FX ratesCommodity cycles: Technicals and fundamentals coincideThe proper use of Gann AnglesSetting stop-losses using price volatility

INTERVIEWRami Habib, Quantitative analyst, Fimat International

SUBJECT MATTERS Identifying commodity currencies

SOFTWAREDresdner Kleinwort Wasserstein's PIA

BOOK REVIEWMechanical Trading Systemsby Richard L. Weissman

COMMITMENTS OF TRADERS REPORTLONG-TERM TECHNICALSEVENTS

0406

081012

1518212427

33

36

39

43

444648

CONTENTS 2 REGULARS>

24 33

Gary Stafford explains how Gann anglesshould be properly applied.

“Volatility arbitrage fund managers prob-ably have around $3 billion under man-agement globally.” - Rami Habib, Fimat.

Equis International has introducedMetaStock Pro FX, a real-time ver-sion of MetaStock specifically creat-ed for FX trading. The packageincludes a FX multiple time framesystem, allowing traders to get a "topdown" view of a FX currency pair,and studies including the FX KeltnerSystem. It looks for patterns thatoccur on a very infrequent basis andinforms the trader of subsequentprice moves. Keltner System 2serves the same purpose but for pat-terns that occur on a more frequentbasis. Also included is the FXPattern System which combinesBollinger Bands, Standard ErrorBands and a Stochastic Oscillator.Interbank FX has developed a pro-prietary execution engine incorporat-ed into the MetaStock Pro FX software allowing traders to deal directly from the charting software.

EQUIS INTRODUCES METASTOCKFX PRO

4 THE TECHNICAL ANALYST July/August 2005

Industry News

LSE to launchfirst oil ETF The LSE will launch the first oil-backed EFT at the end of July. OilSecurities, the company that set upthe first gold-backed ETF inAustralia will probably list two oilsecurities on July 28 in London.Graham Tuckell of Oil Securitiessaid one security would track theprice of the front-month Brent con-tract traded on the IPE while theother would track the West TexasIntermediate contract traded on theNYMEX. He said he was confidentthat the investment in the oil ETFswould surpass the $3.5bn so farinvested in the gold ETFs.

CBOT EXTENDS ELECTRONIC TRADING HOURS The Chicago Board of Trade(CBOT) has announced the expan-sion of trading hours on its elec-tronic platform, a response to globaldemand for additional tradingopportunities at the exchange.CBOT products will be available foran additional hour each day.

The new operating hours for theCBOT electronic platform will be6:00 p.m - 4:00 p.m. (the followingday), with the pre-opening sessionnow beginning at 5:30 p.m.Previously, the hours of operationwere from 7:00 p.m. - 4:00 p.m.,

with the pre-opening beginning at6:30 p.m. The change will be com-pleted in conjunction with theupgrade to the CBOT electronicplatform on October 9, 2005

Industry News


E X C H A N G E N E W S :

IPE energy futures see volume recordafter switch to electronic trading

The Intercontinental Exchange (ICE) say a new volume recordin the benchmark IPE Brent crude oil futures contract was seton July 12. IPE Brent volume reached 185,144 contracts,exceeding the earlier record of 181,182 contracts set onSeptember 14, 2004. ICE say the figures reflect the increase inelectronically traded contracts following the transition in Aprilfrom IPE's open-outcry floor to fully electronic trading.

The CBOT announced that exchange-wide average daily volumeclimbed to a new record in the second quarter of 2005. ADVrose to 2,889,059 contracts, up 14.8% from the same period lastyear. Total volume in the quarter also reached a record184,899,800 contracts, up 16.6% from the same period in 2004.

Euronext.liffe traded a record month in its flagship contract, 3-month Euribor, in June at over 18 million contracts represent-ing an average daily volume of over 834,000. All short terminterest rate products traded a total of 36 million contracts dur-ing June, up 34% on June 2004.

Alan Johnson has joined MacquarieBank in London as consultant AsianTechnical Analyst. Johnson will pro-vide analysis on Asian equity marketsincluding Japan for Macquarie clientssuch as long-term pension funds andhedge funds. Previously he was aconsultant analyst with HSBC and in1987 founded Market DirectionalAnalysis, a research company special-ising in interest rate and AsianPacific equity technical analysis.

ON THE MOVE...

AlanJohnson

ALGORITHMIC TRADINGThe first buy-side survey shows firms have yet to wholeheartedlyembrace algorithmic execution strategies

Special Feature


Although algorithmictrading is a fast devel-oping area within the

financial community, until nowthere has been no detailed dataon the extent of usage of algo-rithmic execution strategies(AES). Financial technologyresearch house, FinancialInsights, and Banc of AmericaSecurities have recently con-ducted the first extensive surveyof algorithmic trading acrossthe trading and investmentcommunity. The results showthat equity buy-side firms haveyet to fully embrace the benefitsoffered by AES.

The use of algorithmic executionstrategies has grown in recent years as aresult of several factors. These includethe decimalisation of security pricesproducing fragmented liquidity acrossmultiple prices which has made achiev-ing best price from execution morecomplex. In addition, recent FSA rec-ommendations for best execution prac-tices have put pressure on fund man-agers to provide a process that ensuresthey are achieving best execution fortheir clients. The market has also seenfalling costs of electronic trading toolsand increased pressure to reduce trans-action costs.

The survey shows that 91% of buy-side firms now use order managementsystems (OMS) in their trading. Use ofelectronic communication networks(ECNs) stands at around 80% of firmswith crossing networks usage at 78.3%

and direct market access (DMA) at73%. The prevalence of these variouselectronic trading practices show thatbuy-side traders are now actively work-ing their orders in the market in orderto achieve lower commission rates andreduced information leakage. Above all,the increased used of electronic tradingtools means that transaction costs arenow seen by fund managers as a signif-icant source of negative portfolio per-formance.

Financial Insights surveyed 60 headequity traders from the top 477 invest-ment management firms, pensionfunds and hedge funds in the USregarding specifically their use of algo-rithmic execution strategies. The surveyshows that on the buy-side 66% of reg-istered investment advisors (RIAs) and92% of hedge funds use algorithmictrading to some extent (see Chart 1).Nevertheless, these figures belie thereal extent to which AES are beingdeployed as 70% of those surveyed saythey are doing less than 10% of their

trading using algorithms. However,usage increases sharply for larger firmswith $70 billion and above under man-agement - 30% of large firms and33.4% of hedge funds of this sizereported algorithmic trading usageabove 10% of total order flow.

The main reason given by buy-sidefirms for not adopting algorithmictrading was that it does not supporttheir fiduciary responsibility for bestexecution. This says Rob Flatley, co-head of sales for electronic tradingservices at Banc of America Securitiesin New York, illustrates that the appli-cation and methodology of algorithmictrading is not yet fully understood.

Benefits of algorithmic trading The survey shows that the buy-sideperception is that the benefit of algo-rithmic trading is in increased produc-tivity rather than improved trading per-formance - head traders say they arehappy to send their easiest orders toalgorithms to get them out of the way

RIA

Hedge funds

Chart 1. 66.7% of firms say they use algorithmic trading to some extent

so they can focus on those that aremore difficult. According to Flatley, agreater focus on transaction costsanalysis (TCA) will lead to the realisa-tion that a more consistent trading per-formance is also a significant benefitfrom adopting algorithmic trading.Hedge funds have led the way in theuse of algorithmic trading systemslargely because they have had more togain from increased productivity andthere is a greater need amongst thesmaller companies to reduce staffcosts. Generally, low commission costsand anonymity are seen as the mainbenefits of algorithmic trading on the

buy-side (see Chart 3). However, thesesame respondents don't see reducedimplicit trading costs (market impactplus opportunity cost) or the consistentout performance of algorithms overtraders as key. Banc of AmericaSecurities stress that these are two ofthe most important benefits of algo-rithmic trading. At present, only 57%of firms surveyed said they expect theiruse of algorithmic trading to grow overthe next 2 years.

Even with those firms that regularlyuse AES, Flatley says the full benefitsoffered by algorithmic trading have yetto be fully exploited. "The majority of

algorithmic trading is executed via themost basic algorithms available" hesays. "Moreover, at only 5% of totalorder flow, AES use is presently lessthan that suggested by recent financialmedia coverage. As such, the scope forgrowth and development of AES use isenormous". For example, the surveyshows that only around 40% of firmsare customizing their algorithmic trad-ing models to suit their own tradingeven though system vendors usuallyprovide customisation at no extra cost.

The sell side The development of algorithmic trad-ing was, in part, due to a response bythe sell-side to pressure on commissionrates from fund managers.Consequently, algorithmic trading hasbecome integral with around 30-35%of trading volume at sell-side bulgebracket firms now executed using algo-rithms. This has led to reduced costswith the ability to scale volume - for thesame costs, more orders can be tradedand desk head counts can be reduced.

Rob Flatley concludes, "Algorithmictrading will extend into futures, foreignexchange and Treasuries in the not toodistant future. It is feasible that algo-rithmic trading volume could doubleeach year over the next two years. Bancof America Securities expects the levelof algorithmic trading order flow adop-tion to increase to 20% by 2007 withthe use of more sophisticated strate-gies. The key for traders is determiningthe best algorithm and its appropriatesettings. This should lead to a moreeven usage across equity market capital-isations. The challenge for the sell-sideis to invest in quantitative resources andsupporting technologies to improvestrategy performance for less liquidstocks."

Further information can beobtained from Sean Mogle [email protected]

Special Feature


50

55

60

65

70

75

80

Lowers

commission rate

Provides

increased

anonymity

Convenience Higher trader

productivity

Provides the

ability to scale

without increasing

headcount

Reduces explicit

trading costs

Reduces implicit

trading costs

Yields more

consistent

performance than

traders

Chart 2. Benefits of using algorithmic trading (% of firms)

Chart 3. Algorithmic trading as % of total trading (% of firms)

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

Don't know >50% 26-50% 11-25% 5-10% 1-4%

Market Views


Perseverance is the word that best describes the trendin gold as we press through the hot summer months.Originating from the cyclical lows of $252, this

metal's spectacular performance relative to major currencies,equities and bonds, has heightened its portfolio appeal with-in the investment community. That said, after registering asixteen-year high at $456 in late December, we are currentlyseeing a period of tempered consolidation. Ultimately, thisyear's tremendous rally in the US dollar, compounded byacute euro weakness (as EU economic growth and unityissues loom), continues to hamper gold on the upside.Nevertheless, the easing of the inverse correlation between

gold and the USD in recent months provides welcome trac-tion for speculative gold bugs and gives gold the opportuni-ty to push on higher.

Price targetsThe seeds of change were planted in 1999, a time whengold ended its twenty year secular decline and started rising.In that year the US stock market was booming and the dol-lar had appeared to be the universal store of wealth for the21st century, while gold had almost become a relic.Following its price slump at the cyclical lows of $252, adouble bottom formed. This base would later dominate the

positive pattern of higher peaks and troughs.Furthermore, gold's bull-phase becamestrongly guided by the rising long-term 70-week moving average (Figure 1). By lateDecember 2004 it had retraced almost onehalf of what had been lost and - althoughtrend followers made further linear projec-tions upwards - momentum then began towane.

In practice, market swings remain complexas bulls and bears battle sideways within aconverging-line pattern. Confirmation above$446/$456 promises to slice through our tri-angle pattern objective at $483, yieldingaccelerated swings towards the December1987 peak at $503, equivalent to the impul-sive 5th wave of an Elliott Wave structure.Settlement below the $410 level would putthis scenario on hold and would providedownside targets at the 25% or 38.2%Fibonacci retracement levels (of the primarytrend), before regaining composure.Assessing the real value of gold against othercurrencies adds further weight to the long-term bullish gold theme. Figure 2 shows theprice of gold in terms of euros, yen and ster-ling all developing positive patterns of higherpeaks and troughs above their rising long-term 200-day moving averages, a widelywatched moving average. Interestingly, closerexamination of the EUR/Gold chart showsthere has been a sustained monthly breakoutfrom a seventeen year-old resistance area.Admittedly, this could be a short-lived emo-tional reaction to EU unity and growth chal-

GOLD:LONG-TERM TREND WILL HOLD by Ron William

Figure 1. Weekly gold chart guided by the long-term 70 week moving average.

Figure 2. Daily gold priced in euros, yen and sterling. Note rising trends above their 200-day MA.


Market Views

lenges. However, while EUR/Gold machinates above €352the risks to the upside remain. Technical measurementsfavour €405 as a minimum upside target (€352 + €53).

Gold/USD correlationFigure 3 shows that the inverse relationship between goldand the USD index (trade weighted) follows a cyclical pat-tern, and looking at the 1-month annual rolling correlationtells us that it is beginning to weaken once again. While thepast couple of years have seen a fall in the USD indexmatched almost exactly by an advance in gold prices, inrecent months this trading relationship has eased from -0.9to -0.7, a reflection of the fact that while EUR/USDdescended to $1.20 from just below $1.25, gold managed toturn upwards towards the $443 level.

The loosening relationship between these two powerfulasset-classes provides welcome traction for speculative goldbugs. Moreover, there is growing confidence that withprices being increasingly driven by fund managers carvingout a bigger place for gold within their porfolios, the dragfrom USD price action can be overcome. If this is the case,then it provides another reason for seeing gold advancingback to the $456, and possibly onwards to the $500/503barrier.

Watch the non-commercialsWhether you are following gold's positive pattern of higherpeaks and troughs or are simply interested in playing thebreak, it is always worth identifying trends in market posi-tioning. To do this, we take a look at the Commitment ofTraders Report (COT). Founded by the CommodityFutures Trading Commission (CFTC), it requirestraders/investors to report their daily positions. The CFTCclassifies traders into three groups: commercial traders(hedgers), non-commercial traders (large speculators), andsmall traders (small speculators). Most TA textbooks rec-ommend watching the commercials, as larger hedgers aretraditionally assumed to be the smart money. However, thishas not been the case with gold over the last five years.Indeed, it has been heavily sold short by the commercialsever since the new bull phase started. With similar negativepositioning against the Euro, commercial sentiment hasfavoured the safe haven of the Swiss Franc.

Instead, the renaissance of gold was helped by dimin-ished sales on the part of central banks, reduced downsidehedging activity by producers and a lack of returns on equi-ties post the bursting of the tech bubble, thus marking thebeginning of the speculative long position on the currentgold up-trend. Figure 4 shows that net-long speculativepositions have successfully tracked the resurgence of gold.Following a record acceleration of speculative activity into2005, positions have since been scaled back. However, cur-rent sizable speculative long positions coupled with a likelyincrease in physical demand (driven by seasonal summertrends), should alert us to renewed upside impetus. (Notegold demand cycles are historically stronger during the sum-mer period when wholesalers typically build up inventoryfor the end of year retail Christmas pick up).

ConclusionThe major upward trend in gold remains dominant, withany short to medium-term weakness indicative of consoli-dation. These corrections should present longer-terminvestors with good buying opportunities. In price terms,confirmation above $456 promises to slice through our pat-tern objective at $483, yielding accelerated swings towardsthe December 1987 peak at $503. Indeed, with oil drivingthe USD, gold should benefit from any medium-term weak-ening of the USD. This effect is supplemented by theincrease in gold demand from the Middle East, as thestrength in oil prices stokes demand through wealth cre-ation. Moreover, persistent underperformance of otherasset classes, coupled with the generalized increase in overallcommodity demand from emerging markets, looks set toentice gold back above the peak levels of $850 in thelonger-term.

Ron William is technical strategist at IDEAglobal inLondon.

Figure 4. Illustrates how net-long speculative positions have been oneof the primary drivers for gold. Note sizeable long positions remain.

Figure 3. Rolling annual correlation coefficient showing the inverserelationship between gold and USD index following a cyclical pattern.Note weakeness in recent months.

FTSE 100:FIRST FIBONACCI TARGET AT 5450by Gökhan Erem

Market Views


In the second quarter of 2005 stock markets showed ustheir positive side again and for many market partici-pants it was an unpleasant surprise. Price action in the

equity markets now reminds me of the second half of thenineties when we saw markets extending without any obvi-ous reason. More and more people were, and now again are,increasingly worried over fundamental issues like the oilprice, the EUR/USD exchange rate and what MrGreenspan is going to do and say at the next FOMC meet-ing. Yet equity markets continue to rise beyond fundamen-tally-based reasoning. And today, market participants aredisplaying the same stubborn unwillingness to change theirmind set and their position in the market.

A more technical description of recent developmentswould be that short interest in the markets has risen toomuch to be able to yield serious profits on downward

focussing positions. In this environment the reaction to anydecline is often short term profit taking or loss limiting buyorders that immediately provide demand and thus takeprices back to the highs and sometimes even beyond thoserecent extremes.

Long-term target (Figure 1)The major trend has been up since the lows set in 2003, butlet's focus on the second half of this trend where the trendwas finally confirmed.

In 2005, we saw the first serious test of the major trend.Prior to this test, the trendline had been drawn under twopoints making it a trendline under construction at most.The decline came and brought prices to the 61.8% retrace-ment level of the most recent rise prior to the decline. Thatsupport held and the trend was fact for the first time.

Figure 1.


Market Views

We can now use Fibonacci levels to look for the next tar-get within this bullish trend: The rise prior to the declinestarted at 4283 and reached a maximum of 5077. After that,the FTSE corrected to a Fibonacci level at 4773. If wemake a 100% projection of the rise from 4283 to 5077(about 800 points) and add it to the new support at 4773,the target becomes 5573.

This target remains valid as long as the stop-loss justbelow the trendline at 4900 remains untouched. Below thatlevel the trend is broken but because that would mean let-ting a lot of profits evaporate before changing direction, I'dplace a stop just below the last top in the trend, in this case5050.

Short-term target (Figure 2)The decline on the 7th of July following the Londonbombings started at the top end of a short term uptrend.The rise was already developing characteristics of an expo-nentially rising trend, suggesting the move was becomingoverextended and unsustainable. At least it needed a correc-tional decline in the short term. So the decline came andtook prices through the short term uptrend until the 50%retracement level was reached.

As on the longer term outlook where the natural reactionlows were formed around Fibonacci levels, the 50% retrace-

ment was met and provided support. In Figure 2 this isillustrated by the use of a Speed/Resistance fan where themiddle diagonal is the 50% retracement line. After measur-ing the price action from the bottom to the end of thetrend, a projection for future prices can be made. The topat that moment was formed at 5225 and the starting pointwas just below 4773, which leaves a move of at least 450points. When we project these points upwards from the5022 level, where prices stopped falling on the 7th of July, atarget of 5450 is found.

Once again, support is found at Fibonacci levels and asecond target can be calculated using Fibonacci projections.This conjuction of two targets increases the probability ofthe target being met.

From a safety point of view, I choose the lowest and thusthe easiest target to reach. This does not mean that the risein prices ends there but implies a high probability of thattarget being met before correcting severely.

To summarize: As long as the uptrend in the FTSE 100,currently at 4900, doesn't give way, the first target for thisuptrend is the 5450 level.

Gökhan Erem is senior technical analyst with theHedge Fund Team at Rabo Securities.

Figure 2.

Since the Fed started raising interest rates on June 30th2004, the logical conclusion must have been that long-term interest rates are due to go higher. To the sur-

prise of almost all market participants they have not. TheFed Chairman has talked about a "conundrum " and manyresearch houses have thrown in the towel and stopped pre-dicting higher long-term rates ahead. Nevertheless, awayfrom the fundamentals, DeMark indicators have proved tobe highly reliable in forecasting the direction of longer-termrates.

The second quarter of 2005 produced a very rare signal -the first quarterly TD Combo sell signal since the Bundfutures started trading in 1990. The signal implies signifi-cant long-term market exhaustion and presents an especiallygood trading opportunity.

Tom DeMark has frequently stressed that the best tradesusually appear when different time periods give similar sig-nals and when similar products confirm. Therefore the keycomponents in judging the quality of the quarterly sell sig-nal are the other time frames as well as the situation inother related interest rate futures markets.

Monthly CountdownFigure 1 shows that the monthly Bund chart gave a "13"TDAC exhaustion indication in May. This is a rare signal inthe 10-year German government debt market. There aretwo prior instances when it timed a market turn extremelywell. The first time was in December 1993 when it appearedin the 10-year yield chart (Figure 2) and the second timewas in December 1998 when it appeared in the futures mar-ket (Figure 3). The chart in Figure 1 highlights the "powerof the 13". In both cases, the market sold off by more than200 basis points in yield terms within one year. In terms ofthe fixed income market this is a crash. A monthly TDCombo exhaustion signal which appeared in June also con-firms the massive corrective potential of the current setup.

Weekly sell signalA look at the weekly chart in Figure 4 displays an interest-ing situation. A TD Combo "13" sell signal appeared onMay 13. Simultaneously a new Setup had started countingand had already reached a "6". In such situations the recom-mended strategy is to wait for a completed Setup or a flip inthe opposite direction, i.e. a closing below the close fourweeks ago.

That strategy helped to avoid getting into the trade too

Market Views


BUNDS: DEMARK INDICATORS WARN OF MAJOR REVERSALby Gordon Kolling

Figure 1.

Figure 2.

Figure 3.


Market Views

early. Three weeks later - on June 3rd - the 13(9) signal wasthere. Looking at the past, such signals are reliable weeklyturning points. Such signals appeared at the bottoms in1994 and 2000 and at the top in 2003.

Daily situationOn June 12 we received a daily 9-13-9 sell signal thatproved very accurate in timing the expected move lower.The market sold off from 123.62 to exactly 121.28. At thatpoint four signals occurred that warned about the quality ofthe down move. 121.28 is exactly the TDST level and abreak below would have been disqualified on that day.

Simultaneously T-Note and Bund futures both had hourly"13" buy signals and most importantly the market had soldoff from the top for three consecutive days - often a reli-able sign that the move is prone to reversal. Consequentlythe market traded back up and even made a new high.When the market spiked up after the London bombings wereceived a new sell signal on an intra day and weekly basis.The new highs have produced evidence that the market isready for a swift reaction lower.

Downside potentialIn the past, reactions to monthly TDAC signals led the →→

DeMark glossary

The TD Combo The TD Combo consists of two distinct stages that are designed to anticipate trend reversals: Setup and Countdown.

Setup: The initial setup phase consists of a series of at least nine consecutive closes less than the close four trading barsearlier for a buy setup and at least nine consecutive closes greater than the close four trading bars earlier for a sell setup.The Setup establishes the environment for the market and determines whether a trader should be looking to buy or sellthe market.

Countdown: Once a Setup is complete, the TD Combo refers back to day 1 of the Setup and begins a Countdown onthat day. The key elements of a sell Countdown are threefold:1. A series of 13 successive closes less than or equal to the low two price bars earlier 2. Each countdown day's high is greater than the previous trading day's high3. Each successive Countdown day's close is greater than the previous Countday day's close and the previous trading day's

close

The TD Aggressive Combo (Version 2) TDAC has lesser requirements for Countdown Bars 11, 12, and 13. They only need to close successively higher for a SellCountdown and successively lower for a Buy Countdown. The aggressive version of the TD Combo is designed to fore-warn of an impeding exhaustion of the trend and often signals a counter-trend correction. Usually the market continuesafter such a correction and completes a Countdown 13 in the non-aggressive TD Combo version.

The TD Termination Count The Count allows an even looser definition for Countdown Bar 13, only requiring that its opening or closing price be high-er than the closing price of Countdown Bar 12.

"Recycling" Recycling is a concern that could arise in a strongly trending market. Generally speaking, if a subsequent Setup occurs priorto completion of countdown, then a new countdown process must begin.

TD Setup Trend (TDST)These are important support and resistance levels. The TDST is defined as the highest price of bars 1 to 9 in a Buy Setupor the lowest price of bars 1 to 9 in a Sell Setup. Furthermore, there are certain rules that are used to determine whethera break through these support or resistance levels should be followed through or if they should be treated cautiously, i.e.false breaks. If the DeMark criteria are met (not discussed here), then the break is qualified. If the criteria are not met,then the break is disqualified (false). See page 15.

The TD Combo and the Bund marketOver the last few years, combining the TDAC and the TD Combo has proved to be a very accurate method for timingcorrections to the Bund futures market. In Figure 2 you can see two TD Combo "13" signals - one in the second quarterof 2003 and one in Q2 2005. Both signals were preceded by a marked downside correction. The size of the correctionwould have stopped out many long positions. This is where the TDAC is of outstanding value. In Figure 1, I have markedout the signals created by the TDAC. It shows how well they predicted the corrections. According to DeMark, correctionsfollowing a TDAC signal usually last for three to 5 periods. I have examined the results for Bunds and found that this rulecan be fully applied to Bunds. Interestingly the market tends to correct 3-5% after such a signal. I suggest using short-termsignals (hourly charts etc) and qualification criteria to time the correction exactly.

Market Views


market back towards the monthly TDST level. This is locat-ed at 114.19. There is a good probability that we shall tradetowards that level in the next six to twelve months. Tojudge the end of the down move, we need to watch pricebehaviour at retracement and TDST levels. The first dailyTDST is at 121.28 and the first weekly TDST is at 118.59.Qualified breaks of such levels are good indications that themove is not exhausted.

ConclusionWhen looking at the combination of quarterly and monthly13 signals in the Bund futures and related markets, the pos-sibility of a major turn in the market must be taken serious-ly.

DeMark indicators are an extremely powerful tool in tim-ing trading decisions. The signals currently in place warn allbond bulls to be very cautious. With quarterly, monthly andweekly sell signals in place the chances are that we are goingto see a significant move lower. The situation compares inmany ways to the two big Bund sell offs in 1994 and 1999. Iam confident that Tom DeMark's indicators will guide usthrough this process with impressive reliability and will helpus to time the down move with the same accuracy as wasshown on the way up.

Gordon Kolling is technical analyst at CommerzbankGroup Treasury and Research.

Sources: CQG; "New Market Timing Techniques" by TomDeMark, John Wiley & Sons, 1997.

Figure 4.

Techniques


Tom DeMark, the wizard's wiz-ard*, is probably best knownfor his TD Sequential and TD

Combo patterns but two others thathave impressed me are TD Lines andTD Absolute Retracements. TD Linesbecause they really made me thinkabout how I was drawing my trendlinesand offer a new perspective on whatconstitutes a proper line break. TDAbsolute Retracements are a new spinon Fibonacci retracements which havespectacularly called some major marketturns.

TD LinesWe know that some lines are moreimportant than others: the more pointstouched, the longer the line, the lessacute the angle, the more important aline is typically considered. The breakof a line that has been tested manytimes and remained unbroken for along time and is not climbing sharply isnormally considered an important linebreak and a big move is expected. ButTom Demark says we have missed thepoint. He says we have got it wrong.

There are many problems with line

drawing. Two technical analysts maywell draw differing lines. Some consid-er a small line break allowable. Othersstrictly join rising lows to draw anuptrend and falling highs to draw adowntrend. There are no 'rules' fordrawing a line. Further, technicians donot agree on what constitutes a linebreak. Some say a one-tick movementthrough the lines is a break, others aclose through the line, others a percent-age movement through the line and soon. TD Lines have strict rules for theirconstruction, importance, the tar- →→

DEMARK RETRACEMENTS AND TRENDLINESby Trevor Neil

Trevor Neil and Tom DeMark

DeMark takes the guesswork out of judging the most important support and resistance levels.

Techniques


get and when a line has had a goodbreak and when the break is likely to befalse. This means that a computer candraw them, set targets when they arebroken and indicate if the break is agood one that will result in a trendchange or a 'false break' which mayeven be tradable against the directionof the break itself.

Line constructionInstead of drawing lines from left toright, down across highs or up joininglows, Tom DeMark suggests lookingfrom the right for a TD Point to thenext TD Point to the left then extendthe line to the right. TD Points are thekey. A one-point TD Point would be abar which has a low below the low ofthe prior bar and the subsequent bar.Imagine a support point that is so pow-erful that is can stop a fall in the marketwhen it has gone on for four bars andcaused a rally which extends for fourbars. This is a powerful support point.Of course if a fall of say six bars is bro-ken and a rally of six bars follows, thisis an even more important supportpoint. TD Uptrend Lines are drawn byfinding the next TD Point to the left

with the same number of falling barsfollowed by rising bars. TDDowntrends are drawn by joining twoTD Points constructed using the samenumber of higher highs followed bythe same number of lower highs. Hispoint is that the shape of the pivotpoint that matters in deciding howimportant an uptrend or downtrendline is. The more highs and lows sur-rounding the pivot point the more sig-nificant the resistance point. Thismeans a line can be drawn using justtwo pivot points because the pivotpoints themselves are so significant.

Setting targetsOnce a line is broken a target is set. Thetarget is calculated by taking the maxi-mum excursion in the direction of thetrend and measuring that amount fromthe break point. The idea is the amountof overexcitement during the life ofthe line will be reflected by proportion-ate disappointment after the break. Soif the line has been hugged by the mar-ket during its life there will be a mildreaction after the break. But if the mar-ket got so excited during the uptrendthat traders pushed the market far away

from the line itself, traders are then lefthigh and dry by the break and will be ina rush to get out of positions that arenow well away from the market. Thiswill cause a big move after the break.

Qualified and Unqualified linebreaksConventional technical analysis hassome difficulty in determining when aline is actually broken and the market ischanging direction. There are too many'false breaks', an innocent name for alosing trade. Tom DeMark gives rulesfrom which it can be determinedwhether the break is for real and isgong to result in a profitable trade. Byfirst qualifying a TD Line, the traderreduces the risk that intra-day entry(going with the TD Line Breakout) willfail, and, at the same time, increases thechance of success should the user fadethe disqualified trade (going against theTD Line Breakout). To be a QualifiedTD Line, price must satisfy one of thefollowing conditions:1. The close of the bar prior toupside/downside penetration isbelow/above the close two bars ago,OR 2. The open of the breakout bar isabove/below the TD Line and alsoopens above/below the previous bar'sclose, OR 3. The difference between the previousbar's close and its true low/high addedto/subtracted from the previous bar'sclose is less than/greater than the TDLine. A true high is that bar's high orthe close of the previous bar, whichev-er is greater; a true low is that bar's lowor the close of the previous bar,whichever is less.

To be a Disqualified TD Line(dashed line), NONE of the aboveconditions exist, and the reverse logicapplies.

TD Lines give traders: an exact a reli-able way to draw lines using significantsupport and resistance lines; automatictargets following the break; and anautomated determination of the validi-ty of the break itself.

Figure 1. TD Lines on EUR/USD. The solid lines are Qualified and when they are broken a tar-get is set. Note how on the break of 11/04/05 at 1.2940 an automatic target of 1.31 was set andreached a few days later. Note also that the line breaks on June 1st and 14th were false. The lineitself went dashed indicating it was Un-Qualified. This happened on the day of the break andtold us it was false. A short term counter trendline break trade could have been taken at thesepoints. TD Lines can be used on intra-day charts for more trades or weekly charts for a longerterm perspective. All lines are drawn by the software package.


Techniques

Absolute RetracementsJust as there are subjective problems indrawing trend lines, there are also thesame problems when deciding how tomeasure Fibonacci retracements. Theproblem is choosing the measuringpoints. Where to start looking for the38.2%, 50% and 61.8% levels? Weknow these levels can be very effectivebut which high or low is the right oneto measure from? Tom demark offersus an objective way to make thesemeasurements which does not requireus to guess which price level we arereacting from.

Tom DeMark suggests taking one ofthe two knowns: the all-time high or all-time low and measuring down to zeroor up to the high and dividing theseinto 31.2%, 50% and 61.8% levels.

TD Absolute Retracement has animpressive record of finding importantsupport and resistance levels that holdsignificance for many years. All thiswithout the trader having to guess thestarting point for the Fibonacci meas-urement.

Tom DeMark with his TD Lines andTD Absolute Retracements has givenus two mechanical and objective meth-ods of describing trends, determiningwhen they are broken, setting targets,and a way to find important supportand resistance levels without having tomake subjective judgments over thestarting point for measurement.

Trevor Neil is a hedge fund manag-er at T-Capital in Cape Town. Hewas head of technical analysis atBloomberg for 4 years before mov-ing to South Africa to form a newtechnically traded fund. He offersinstitutional training through iscompany BETA Group.www.betagroup.co.uk.

* Jack Schwager

Reference: TD Lines and TD Retracements, pleaserefer to Thomas R. DeMark, The New Science ofTechnical Analysis, New York: John Wiley & Sons,1994.

Figure 2. TD Absolute Retracement off of the 24/03/00 S&P500 cash high day's close. TheTD Absolute Retracment levels downside are 944.75 and 764. The 944.74 was the exact 9/21/01low and the 22/10/02 low fell 2 points short of the objective. See X's on chart.

Figure 3. TD Absolute Retracement upside from the close of the day after the October 2002low. A rally to the 150% TD Absolute Retracment objective at around 1252 would be equivalentto the rally subsequent to the April 1930 DJIA rally following the 1929 crash and prior to thehuge fall to the 1932 low. That would be comparable to the S&P subsequently declining by halfbut very possibly by 61.8%.

Techniques


INTERMARKET ANALYSIS:FORECASTING SHORT-TERM FX RATESby Marc Zaffran

The forward and futures markets canalso provide a good indicator of ratesfor the purposes of cor-porate planning. However,their use should not beconfused with mistakennotions about the accuracyof their predictions. Afterall, market rates are deter-mined by the bestinformed participants andby those participants whoare able to move first, notby market consensus.

To provide a betterinsight into future marketbehaviour, the SocieteGenerale Corporate andInvestment Banking FX &Derivatives Sales Team inLondon perform analysescalled "Market Behaviour& Historical Pattern"(MBHP).

Looking for parallelsCorporate and institutional clients areconstantly asking the sales desk ifrecent events in the market are similar

to events that have happened beforeand, if they have happened before,what was the impact on the FX marketsthat followed.

To provide an answer to these ques-tions, our MBHP analyses essentiallyask two questions of our proprietarydatabase: Can we find historical paral-lels for recent significant events in themarkets? If so, what happened next?

The analyses are executed weekly andare based on events in stocks, bonds,commodities, energy, and foreignexchange markets, and sometimes eveneconomic and political data such as US

payroll figures and general elections.With this database, which holds data

going back a minimum of 35 years, wecan ask questions such as "when priceswent up in x market and down in y mar-

ket over a z time period, what hap-pened to ab currency pair in the follow-ing week?" The database is sufficientlydeep and sophisticated in its formula-tion to allow us to ask almost any per-mutation of question, although limitingourselves to a maximum of three tofour parameters tends to yield the mostreliable results.

We generally look for, and publish,results that suggest a 70% or greaterbias in behaviour. Of the 40 weeklyanalyses we have published in the lastyear, 29 have been successful.

CorrelationThe questions we poseour database are basedon an understanding oreven a hunch about therelationships betweenmarkets. Thus perform-ing historical analysisleads us to highlightinter and intra FX mar-ket correlations.

Correlation does notimply causation in anyway. In other words, justbecause two events arecorrelated does notmean that one causesanother, or has anythingto do with the other.Correlations deal onlywith observed instancesof events. However, a

strong correlation does often warrantfurther investigation to determine cau-sation. It is important to understandhow different currency pairs move inrelation to each other.

Corporates and institu-tions often seek waysto predict currency

rates in order to decide whento hedge or when to take aposition in the market. Themethods they use are typical-ly based on an assessment offundamental or technicalanalysis.

Techniques


Having this knowledge allows you toeffectively diversify and manage yourportfolios. Furthermore, we can tellfrom our dynamic correlation chartshow a correlation between two under-lings evolves through time, providingimportant information about thevolatility of the correlation and ofwhether the current correlation coeffi-cient is reflective of a recent trend orwhether it is fundamentally high.

Historical analysisEssentially, the goal is to identify con-sistent and reliable patterns (or incon-sistent ones) - based on history. Manycurrencies don't exhibit any particularpatterns of movement and this is infor-mation that is useful in itself, but forthose that do our analysis may providesome powerful trading opportunities.

MBHP allows us to investigate thepast in a flexible and efficient way,drawing patterns in the market that arelikely to occur again today. Eventhough the analysis is not usually statis-tically significant - the number ofoccurrences we are working with areoften as small as seven - MBHP has agood track record of highlighting situa-tions which confirm the trend or indi-cate its imminent reversal.

Marc Zaffran, FX & DerivativesSales team, Societe GeneraleCorporate & Investment Banking,London.

1. In the one week from the 22ndOctober 2004 to the 29th October2004:

EUR/USD gained more than 1% The Dow Jones Industrial Averagegained more than 2% (269 points) NYMEX Light Crude Oil futurelost more than 5%

We asked our historical database howEUR/USD, CABLE and USD/CHFhad performed WHEN:

EUR/USD had been up over theweek &NYMEX Oil Futures had lostmore than 5% over the same week&The DJIA30 had gained more than2% over the last three trading days?

We found 17 prior occurrences of thisscenario since 1986 and historyshowed that after these events:

EUR/USD rallied in 82% of thecases with an average of 1.90%over the next 10 trading days.The bearish signal for the dollarwas confirmed on other currencypairs:GBP/USD rallied in 70% of thecases by an average of 1.40% overthe next 10 trading days.USD/CHF decreased in 82% ofthe cases with an average of 1.70%over the next 10 trading days.

The actual result:

EUR/USD gained more than1.40% from 29/10/04 to12/11/04 on a closing bid basis.

2. In early May one of our clients witha significant exposure to aluminiumasked us to perform an analysis thatwould explore the co-movement

between EUR/USD and aluminium.Worried about its short termEUR/USD and aluminium exposure,our client wanted us to determinewhether or not EUR/USD wouldcontinue its two week downwardtrend. Over the same period we saw acorrection in aluminium's six monthsbull-trend with contracts trading onthe London Metals Exchange belowtheir 200-day moving average. Hadwe seen such a simultaneous move inEUR/USD and aluminium before?

We asked our historical database:How had USD performed against theEuro and the Yen WHEN:

EUR/USD had been going downover the previous 10 days & LME Aluminium prices were upbut below the 200-day movingaverage The US dollar had risen in eachof the prior four trading days

We found 8 prior occurrences (from1993 to today) and history showedthat:

EUR/USD lost in 7 cases out of8 by an average of 1.00% overthe next 10 trading days and USD/JPY rallied in 7 cases out of8 by an average of 1.10% overthe next 10 trading days.

The actual result:

While many market participantswere expecting EUR/USD tohead back up, this study allowedus to tell our client that based onthe historical Aluminium-EUR/USD relationship, theEUR/USD downtrend would goon.EUR/USD lost more than 2.00%from 05/05/05 to19/05/05 on aclosing bid basis.

•

•

•

•

•

•

•

•

•

•

•

•

•

•

•

•

•

Two illustrations of MBHP“IT IS IMPORTANTTO UNDERSTANDHOW DIFFERENTCURRENCY PAIRS

MOVE IN RELATIONTO EACH OTHER.”

- MARC ZAFFRAN

Asia and Japan Hedge Fund Directory

Qsjodjqbm!Tqpotps;!

Tfdpoebsz!Tqpotps;!

2005

EurekaProductAD220405.pdf 19/05/2005 12:18:12

Techniques


Empirical studies suggest that a softasset to hard asset rotation typicallylasts about 12-15 years and with the lowin hard assets possibly coming in on the1998-1999 low in the CRB Index at theearliest it is possible that generally ris-

ing commodity prices could be in placeuntil 2010 (see Figure 1). There is alsocompelling evidence to suggest thatboth longer term fundamental andtechnical issues are at work and thatcommodity prices won't becomeexhausted on the upside very soon. Infact, it could be a mistake to discountthe combined impact of the "newmoney" coming into the commoditymarkets and the rampant globalizationof markets in general. Commoditymarkets have never seen the type ofmoney that has begun to flow their wayuntil recently, and they certainly havenever seen the type of demand patternsthat have been developing as a result ofthe rapid economic expansion in China.

In addition to long term technicalcycles (like the Kondratieff wave) turn-ing upward, a number of fundamentalcycles are also positive for commodi-ties. Rising consumption in India andChina alone might present the com-modity markets with the most concen-trated expansion in demand that has

been witnessed in modern times. Infact, considering the opening up ofeconomies that were previously closedto capitalism and their growth in popu-lation since the end of World War II,we suspect that commodity demandincreases could easily outstrip theboom of the 1940's and 1950's.Furthermore, after decades of just-in-time inventory buying and a desire tooptimize profitability, many producerslack the infrastructure to run produc-tion at record levels for an extendedperiod of time. Copper is a good exam-ple and Figure 2 shows how stocks arenow heavily depleted in this metal.Likewise, the oil sector is another primeexample of a commodity market thathas been unable to satisfy worlddemand. Even with significant capitalflowing toward energy investments, it isclear that meeting and exceeding theworld's needs is at best a difficult task.

In some commodity markets, tech-nology has certainly stepped up to helpproduction keep pace with the

Figure 2.Figure 1.

COMMODITY CYCLES:TECHNICALS AND FUNDAMENTALS COINCIDEby Dave Hightower

The commodity marketshave received a lot ofattention from talk

that long term commoditycycles are in the process ofturning higher. There hasbeen so much bullish talkabout commodities that somecontrary players think theMarch 2005 high in the CRBIndex marks the end of theupward cycle. There are anumber of forces comingtogether however that shouldextend commodity prices fur-ther upwards.

→→

Techniques


growth in world demand. However, inmarkets that are subject to the effectsof nature and, in a sense, weathercycles, it is possible that meeting grow-ing demand with ever increasing recordproduction may prove to be too diffi-cult to achieve, creating periodic volatil-ity events. For instance, the grain mar-kets from 1997 to the middle of 2003saw extremely favourable global weath-

er conditions, but those favourable pat-terns seemed to come to an end inAugust 2004 and have continued toworsen into the spring and summer of2005. Therefore, even though geneticengineering and advances in crop tech-nology have resulted in unprecedentedgrowth in crop yields, they are still sus-ceptible to the vagaries of weather.

While the world supply of soybeans,

for example, has been able to sustain anear linear uptrend, world demand hassometimes grown at an exponentialpace. For example, it is estimated that astrikingly large percentage of theincome growth in China is spent onfood and food products. China's soy-bean usage was near 10 million tonnesin 1981, hit 20 million tonnes by 1998and is projected to be 43 million tonnesin 2005.

On top of traditional cycles for com-modity markets, commodity cycles cansometimes be weighted with a naturalcycle which is seasonal in nature. A 91day cycle has been a useful tool fortechnical analysis in the grains and live-stock because production seasons (i.e.spring, summer, fall, winter) appear tohave a significant influence on prices.For example, grain traders watch forsigns of tops or bottoms around March21st, June 21st, September 21st andDecember 21st. The November 2005soybean contract appears to have put ina significant cycle top near June 21,2004 (628), September 22, 2004 (5651/2), December 23, 2004 (569 3/4),March 21, 2005 (606) and the currentcontract highs on June 22, 2005 at 754.

In addition to the historical conflu-ence of technical cycles and rampantglobalization, the commodity marketsare seeing an unprecedented influx of"new money" flowing their way in theform of commodity fund activity (seeFigure 3). This can be seen in the openinterest figures - in 2004 all time highswere achieved for nearly every com-modity (Table 1). In fact, the moneyflowing into commodities since thebeginning of 2003 has the potential toswamp all but the most substantialcommodity markets.

On April 19, 2005 almost every phys-ical commodity market on the boardrallied in a concentrated fashion andreports from the trading floor suggest-ed that a single commodity fund wasresponsible for sparking the buyingwave. Coincidentally, on that same daythe commodity markets saw an unrelat-ed story concerning the launch of theTable 1.

Figure 3..


Techniques

first exchange-traded fund that wasbased on a commodities index.

The importance of this new form ofinstrument and the expanding presenceof fund activity in commodities couldhave a profound influence on howprices are established on all major com-modity exchanges. While anyone canthoroughly document the influx offund money into commodities, the cre-ation of exchange-listed commodityspecific funds could facilitate an evenmore significant flow of money intothe markets, and that in turn couldchange a number of historical assump-tions. In other words, the potential tooverwhelm various commodity marketsis increasing. In fact, in the comingenvironment even if fundament devel-opments prove to be only moderatelybullish or bearish, it is possible that his-torical assumptions of both low andhigh values for specific commoditieswill be redefined constantly and thatcommodity price volatility will becomeexplosive.

Dave Hightower is president of TheHightower Report, specialists infutures research and forecasting(www.futures-research.com).

The Kondratieff Wave represents an attempt to explain patterns of regular,structural changes in the world economy. It is named after NikolaiKondratieff, a Russian economist working in the 1920s, who saw the capi-talist economies as evolving and self-correcting, contradicting the Marxistnotion that capitalism was doomed to collapse. Kondratieff looked at pricesand output in Britain and the US going back to the 1790s. Based on his stud-ies, he determined that economic activity could be measured in "waves" last-ing 50-54 years.

The cycle begins with what is called the "upwave" during which prices startto rise slowly as economic expansion begins. During the course of theexpansion inflation but gradually picks up steam so that after about 25-30years it is running at very strong rate. This sets the economy up for a severerecession that is longer and deeper than anything that the economy hadexperienced during the upwave. Eventually prices stabilize, the economyrecovers and a second wave of expansion begins, but the growth is nothinglike what was seen during the long upwave. This persists for about ten years,but growth is anemic and the economy never reaches its previous dynamicstate. This secondary plateau ends with a sudden shock, such as a financialpanic or a stock market crash, and the economy moves into its secondarycorrection phase, which is far deeper and which leads a depression anddeflation.

There are different theories as to what causes this wave. One theory is thatsome key innovation drives a period of economic realignment and expan-sion, such as the industrial revolution in the 1800s, electronic and motorvehicles in the early 1900's and the information revolution in the latter partof the 20th Century. This realignment either brings about or is necessitatedby the wave action. Another theory is that a new wave begins every threegenerations or after that the group that experienced the previous wave diesoff.

Determining at what phase of the cycle the economy is currently becomesa matter of interpretation. Some students of Kondratieff Wave theory haveput the beginning of the mid-20th Century a wave at 1940 (the end of theGreat Depression and start of the expansion at the beginning of World WarII), while others believe the wave began around 1949 with the post-warexpansion. Using these dates as a point of reference puts the start of thenext upwave at 1990-94 or 1999-2003, respectively. On the other hand, therehas been further theorizing the Kondratieff Wave may have increased inlength to 60 years or more as life spans have increased. Using a 60 year wavelength would move the start of the next upmove to either 2000 or 2009.

Clearly, the timing of the Kondratieff wave is not an exact science.However, what is consistent is that in order for the theory to hold, the mar-ket will have to face some sort of crisis prior to starting the next upmove.Was that crisis the bursting of dot.com bubble or is there something else instore?

Kondratieff Wave

Techniques

THE PROPER USE OF GANN ANGLESby Gary Stafford

Most people proba-bly have softwarethat allows them to

look at Gann Fans or Angles.However it might surpriseyou to find that the softwareis probably doing it com-pletely wrong. What we readregarding the principles fromthe Gann literature availableis not necessarily what isbeing applied.

One of the worst examples I havecome across from a major softwarepackage is where you put a trendlineon a chart and the program then plotsthe other angles based on the originaltrendline. This is a common and funda-mental mistake and bears no resem-blance to what Gann actually said.Gann was looking for geometric rela-tionships between price and time. Thetrendline/Gann fan completely ignoresthis principle.

Even if the software adheres to whatGann literally said you will still beingdoing it wrong in most cases.

As an example, we are going to lookat the FTSE 100 just before the all timehigh in 2000. Figure 1 shows a Gannfan, but based on what Gann literallysaid. As you can see from the chart, Ihave angles spreading out from theOctober low and have drawn the daily1x1, 2x1, 4x1, 8x1, 16x1. However, asyou can see, these angles do not work.

Now we are going to do this correct-ly. Every stock, commodity, currencyetc works in its own unique way, theskill is in finding how it works.

For instance the Dow Jones worksslightly differently to the FTSE 100,the mathematical principles do not

Figure 1.

Figure 2.


change. The FTSE does not work ondaily angles. Gann in his writings calledit "The Master Time Factor". Lookingat Figure 2, the correct angle series isput on using the correct time factor.

The angles Gann talked about are asfollows;

1x1, 2x1, 4x1, 8x1, 16x1, 32x1, 64x1,128x1, 256x1 etc…

In Figure 2 we have put the sameangle progression on but with the cor-rect Master Time Factor. As you cansee the 512x1 angle did not support theprice, the next angle below 512x1 is the256x1 which gives support in February1999. Remember that this angle is not atrendline, it is a mathematically calculat-ed angle which is drawn from theOctober low, almost 5 months inadvance. The next angle drawn fromthe same point is half of the 256x1,which is the 128x1. This gives us thelow in August 1999, the next angle forthe future is the 64x1 which works butis not shown on the chart.

There are other anglesQuoting directly from Gann's writings,he states "After 50 years of research,tests, and practical applications I haveperfected and proven the most impor-tant angles to use." These numbers areshown above, i.e. 1x1, 1x2, 1x4 etc. Theimportant thing here is that they are notthe only angles that work.

Figure 3 shows the angles halfwaybetween the major angles.

The first two steepest angles onFigure 2 are the 512x1 and the 256x1.The halfway angle between these two isthe 384x1 angle. Half of this thenbecomes a 192x1, then 96x1, then 48x1and so on.

Looking at Figure 3 you can see thatthe 384x1 gives us the low in December1999, the 192x1 give us the low in June2000 and the 96x1 gives us the exactlow in October 2000. Now comparethis chart with the angles in Figure 1and you can see how easy it is to getGann angles completely wrong. →→

Techniques


Figure 3.

Figure 4.

“THE UNDERLYING PRINCIPAL BEHIND[GANN]…IS THE FACT THAT SPECULATIVE

MARKETS ARE MATHEMATICAL IN NATURE.”- GARY STAFFORD, GANN MANAGEMENT

Techniques


Another example: GBP/USDFigure 4 is a weekly chart showing the1985 major low, when the dollar almostreached parity. It was from this low wehave drawn a 1x12 angle which gives usthe major low in 1986 and 1987. Thisangle is what we call a half angle, theone between Gann's major angles of1x8 and 1x16. As this angle has workedi.e. supported the price, the next anglewe would watch with great interest istwice the 1x12, which is a 1x24. As youcan see, this angle gives you all the lowsin 1989. The next angle we wouldwatch with interest is the 1x48, whichyou can see in Figure 5.

In Figure 5 we have changed to amonthly chart. As can be seen the 1x48did not support the price, hence thisdid not work. From the logic so far youwould expect us to use the next angletwice 1x48 which would be 1x96.However I found out many years agothat if an angle does not work, it canchange what will work in the future.

To recap, the major angles are:

1x1, 2x1, 4x1, 8x1, 16x1, 32x1, 64x1,128x1, 256x1 etc…

We will call this series 1.

The angles halfway between series 1are:

3x1, 6x1, 12x1, 24x1, 48x1, 96x1,192x1 etc…

This is called series 2.

We can see that GBP/USD has beenworking on series 2. However the 1x48did not work. This means that we willchange to series 1 angles for the future.So the next angle below the 1x48 is the1x64, which is a series 1 angle. Fromthe chart you can see that it gave us theexact low made in May 1996 - this anglewas put on the chart eleven years agofrom the major low in 1985.

The 1x64 has worked so we stick withseries 1. The next angle is therefore the1x128 which is twice 1x64, and youshould be able to see the bottom wasthe exact low in June 2001, 16 yearsfrom the low in 1985.

Mathematical marketsThe underlying principal behind theserules - rules which I have developedover the last 14 years - is the fact thatspeculative markets are mathematical innature. Recognising this fact shouldgive traders a fighting chance in profit-ing from the markets.

Gary Stafford is technical director atGann Management Limited in theUK. (www.gann.co.uk)

Please note there are a series of videosto accompany this article. These can beaccessed online at www.gann.tv (clickon Free Gann Video button)

Figure 5.

“AFTER 50 YEARS OFRESEARCH…

I HAVE PERFECTEDAND PROVEN

THE MOST IMPORTANT ANGLES

TO USE.”- W.D. GANN

Techniques


By accounting forvolatility, the variancein volatility, and

volatility skew, KaseDevStops provide a statisti-cally sound basis upon whichto place stop-loss and stop-and-reverse orders.

The idea of using volatility-basedstops first occurred to me after I readWelles Wilder's New Concepts inTechnical Trading, published in 1978.In this book, Wilder sets out a stop andreverse system centred around a meas-ure called "True Range". It was myreading of that book along with theburst in volatility that took place in theoil market in 1990 that led me to inves-tigate the use of True Range in settingthe size of stops, and ultimately to thedevelopment of the Kase DevStop sys-tem in 1991.

The DevStopA number of trading techniques useaverage True Range (ATR) values withfixed multipliers to determine where toplace stop and reverse orders - forexample, stops might be set at 3 x ATR(see Box for definition of TR andATR). Yet while True Range is propor-tional to volatility and using such anapproach is an improvement over afixed-value stop, the use of the average(ATR) is insufficient to truly capturemarket behaviour.

Think of two populations, PopulationA and Population B. Population A iscomprised of chorus girls and has anaverage height of 5'10" - the shortest is5'9" and the tallest 5'10½". Population

B is comprised of professional basket-ball players and their elementary schoolaged children, and has an averageheight of 5'10" - the shortest is 3'2" andthe tallest 6'11". Obviously, eventhough both populations have the sameaverage height, a higher value is neededto be equivalent to, say, the 95th per-centile.

The same is true for a market. A stopthat is going to perform optimally mustconsider the variability of range, notjust the average. So I developed theDevStop, where "Dev" stands for stan-dard deviation around the mean.

The value of the DevStop is calculatedas follows:

DevStop Amount = average(2 x TR, n) + stddev(2 x TR, n)

For the simple reason that True Rangevalues based on single bars are toosmall, notice the DevStop employs theTrue Range of two bars (2 x TR),

which we call the True Range Double(TRD).

We can use this calculation to deter-mine the points at which the stopsshould be set. For a normal bell curve,for example, a stop placed at the onestandard deviation level has an 85%chance of being hit and a stop placed atthe 1.65 standard deviation level has a95% chance of being hit.

Volatility skewThe problem with using a normal bellcurve is that it doesn't accurately reflecthow range is really distributed. Whenwe look at range as a proxy for volatili-ty (which is a logarithmic term) therange is more or less log normally dis-tributed as opposed to normally distrib-uted. Thus it has a right hand skew.

Figure 1 shows the distribution of thedaily TRD values for the Live Cattlecontract from August 1969 to August2004. It can clearly be seen that there isa severe right hand skew. Indeed a fulltwo-thirds of the data are below

SETTING STOP-LOSSES USING PRICE VOLATILITY by Cynthia A. Kase

True Range

True Range (TR) is the full price range of a period and is defined as the largestvalue of the following three calculations:

1. TR = Highest price of current period("H") - Lowest price of current period("L")

2. TR = Highest price of current period("H") - Closing price ofprevious period("C")

3. TR = Closing price of previous period("C") - Lowest price of current period("L")

Average True Range (ATR) is the simple average of True Range over the past nperiods or an exponential moving average.

Source: The Encyclopedia of Technical Market Indicators, Colby, 2003

→→

Techniques


the average TRD of 1.4 cents.Therefore a correction of the normalstandard deviation levels must be madeto account for the skew.

Back in the early 1990's when we firstdeveloped the DevStop we performedresearch on 100 bar subsets of a totalof 10,000 bars of 10 minute BritishPound data to estimate the skew cor-rections that would be needed to equateto 1.0, 2.0 and 3.0 standard deviations.We found that the one standard devia-tion level, on average, did not require acorrection, and that the two and threestandard deviation levels needed to becorrected by about 10 percent and 20percent respectively, that is, to 2.2 and3.6 standard deviations over the mean.Figure 2 shows an example of theDevStops on the daily chart for theOctober 2004 Lean Hogs contract,where each line has been calculated asfollows:

Warning Line = average(TRD, n) + 0stddev(TRD, n)

Dev 1 = average(TRD, n)+1stddev(TRD, n)

Dev2 = average(TRD, n) + 2.2stddev(TRD, n)

Dev3 = average(TRD, n) + 3.6stddev(TRD, n)

The computer does not know, ofcourse, if one is long or short, so amoving average crossover system isembedded into the front end of thecode. If the fast moving average, nor-mally defaulted to five (5) is above theslow, defaulted to 21, then the comput-er assumes you are long and trails thestop from the highest high since themoving averages crossed. If the fast isbelow the slow, the DevStops trail fromthe lowest low. The stops are flippedfrom below the data to above the datewhen the averages flip from long toshort and vice versa. But the front endcould be programmed for any entrysystem.

Adjusting for skewRecently I decided that it was time toupdate the original study. Rather thanusing 10-minute bars on one contract,we decided to test daily data over arange of physical commodity andfinancial futures from contract incep-tion to September 2004, amounting tomore than 50,000 data points. In addi-tion, we also evaluated the data in 250bar sets which give a more representa-tive population than the 100 bar sets,stepped backward in 100 bar incre-ments. Table 1 lists the study data.

The first question I addressed as Ilooked at the results is how the data

compares with our prior adjustments,i.e.: no correction at the one standarddeviation level, a 10 percent correctionat the two standard deviation level anda 20 percent correction at the threestandard deviation level, resulting inone, 2.2 and 3.6 standard deviationsover the mean for the DevStops. Table2 shows the percentile rankings andnumber of standard deviations for anormal bell curve versus the averageand median values for the actual data.

I found that the warning line, whichis set at the average of the data, or zero(0.0) standard deviations over themean, actually captures about 60 per-cent of the observations. This can beseen by looking across the row marked"60" in bold under the percentile col-umn. Where a normal bell curve wouldrequire a setting of 0.3 standard devia-tions over the mean to be at this point,the actual data only requires a setting ofzero.

At the 85th percentile the one stan-dard deviation setting for normal bellcurve corresponds to about 0.95 forthe actual data, which means that a set-

Figure 1. Distribution of TRD

Table 1. Data used to September 13, 2004

Table 2. Normal Bell Curve vs. Actual Data

Techniques


ting of 1.0 for DevStops is roughlyequivalent to a very slightly higher per-centile reading at about the 86th per-centile. At the 90th percentile, markedin italics, the curves cross from theactual data being at a standard deviationlevel slightly less than that generated bythe bell curve to slightly more. So using1.28 for a DevStop setting would onaverage generate a stop at the 90th per-centile, the same as a bell curve.

At two (2.0) standard deviations, acorrection of 20 percent would beneeded to correspond with the actualdata, the same as has been used allthese years for the three (3.0) standarddeviation stop. Rather than correspon-

ding to 2.0 standard deviations over themean at the 97.5 percentile, a setting of2.2, on average, corresponds withabout a percent lower, that is, with the96.5 percentile. So the use of 2.2 ratherthan 2.4 has been accurate within onepercent of its original intent. As noted,the use of 3.6 - a 20 percent correctionto 3.0 - is almost exactly correct.

Finally, I looked at the degree of vari-ation among the futures contracts stud-ied (Table 3). While a fairly wide rangeof variation between each data set wasfound (each data set being 250 bars),there is little variation among futurescontract. The average standard devia-tion for the results by data set is 0.13

(results not presented here), but theaverage variation by commodity is only0.03, less than 25% that of the variationby data set. This means that the rangeof variation by data set is about thesame no matter what is traded.

In summary, the methodologies thatwere used a dozen years ago to developthe Kase DevStop are still valid. Theone standard deviation level used forDev1 and 2.2 standard deviation levelused for Dev2 are within a percentile ofthe settings that were the originalintent, specifically the 86th vs. 85th per-centile for Dev1 and 96.5 versus 97.5for Dev2. For Dev3, the original settingat 3.6 is right on the mark. There issome variability around the mean val-ues found, but there is insignificantvariation by commodity. That it,regardless of the commodity studied,the results come out about the same, sono adjustment must be made to theDevStops based on the futures contracttraded.

Cynthia Kase is president of Kaseand Company, Inc.(www.kaseco.com). Kase DevStopis available on CQG, TradeStation,eSignal, Aspen Graphics and DTNProphetX.Table 3. Average Standard Deviation Results by Commodity vs. Percentile of Data

Figure 2. DevStops on October 2004 Lean Hogs Daily Chart

“A STOP THAT IS

GOING TO PERFORM

OPTIMALLY MUST

CONSIDER THE

VARIABILITY OF

RANGE.”

CYNTHIA KASE

Technical Analysis in the Commodity and FX Markets

Register online at www.ta-conferences.com

Tuesday August 23rd, 2005Venue: ArabellaSheraton Grand Hotel

Topics covered include:▪ Intermarket analysis ▪ Maximising the value of technical indicators ▪ Using Fibonacci to trade grains ▪ Assessing the rand and gold markets▪ Using DeMark to generate buy & sell signals▪ Commodity markets using candlestick charts

Thursday August 25th, 2005Venue: Intercontinental Sandton Sun & Towers

Topics covered include:▪ Intermarket analysis▪ Maximising the value of technical indicators▪ Using Fibonacci to trade grains ▪ Assessing the rand and gold markets▪ Using DeMark to generate buy & sell signals▪ Commodity markets using candlestick charts

Cape Town, South Africa Johannesburg, South Africa

presents

David Sneddon,CSFB

Pieter Van Wyk,Commodity Investment

Services

Kevin Edgeley,Goldman Sachs

Judy Padayacee,ABSA Bank

Paddy Osborn,TraderMade

Jeremy Goldwyn,Sucden

A series of one day seminars designed to equip traders, analysts,

hedge funds and fund managers with the most effective and up-to-

date charting techniques to trade the commodity and FX markets.

With contributions from leading market analysts, the seminars will

present a range of studies and market views to enhance both short

and long term trading returns.

Register online at www.ta-conferences.com

Tuesday September 27th, 2005Venue: Intercontinental Dubai Hotel

Topics covered include:▪ Trading base metals ▪ Using volatility bands as an overlay on oscillators▪ Assessing the outlook for precious metals & crude ▪ Intermarket analysis ▪ Maximising the value of technical indicators▪ Commodity markets using candlestick charts

Tuesday October 25th, 2005Venue: Trinity House EC3

Topics covered include:▪ Applying behavioural finance to the oil market▪ Using Fibonacci▪ Commodity currencies▪ Maximising the value of technical indicators▪ Trading base metals ▪ Intermarket analysis.

How to Register1. Go to www.ta-conferences.com

2. Call +44 (0)20 7833 1441

3. Email [email protected]

to: [email protected]

Registration fee = £399 (Dubai £599) £100 discount when you register early

Sponsors

Thomas Anthonj,ABN Amro

John Noyce,Citigroup

Shaun Downey,CQG

Robin Griffiths,Rathbones

Trevor Neil,T-Capital

Ron William,IDEAglobal

Dubai, UAE London, UK

Widen your horizons

EUROCAPITAL MARKETSDERIVATIVESSTRUCTUREDFINANCE

Soc

iete

Gen

eral

e is

aut

horis

ed b

yB

anqu

e de

Fra

nce

and

the

Fina

ncia

l Ser

vice

s A

utho

rity,

and

is re

gula

ted

by th

e Fi

nanc

ial S

ervi

ces

Aut

horit

y fo

r co

nduc

t of U

K b

usin

ess.

Choosing a dynamic and innovating bank provides you with new perspectives and widened

horizons, to help you achieve your ambitions today and tomorrow. With SG Corporate

& Investment Banking all our clients - from corporate clients and financial institutions to public

sector clients and investors - benefit from sound analysis, reliable advice and the best financial

solutions. With a growing worldwide leadership in our areas of excellence, we will do our best

to respond to your financing, capital management or investment requirements by combining our

areas of expertise, our innovation and our cross-product approach ■ SG CIB, your partner in Euro

Capital Markets, Derivatives & Structured Finance ■ www.sgcib.com

Interview


TA: Volatility arbitrage remains something of a specializedarea within the hedge fund community. What is Fimat'sinvolvement in the market and your role in the firm?

RH: We are in the Prime Brokerage group of SocieteGenerale where I look after the quantitative analysis ofhedge funds for investors. Our product is the FVAM (FimatVolatility Arbitrage Median) which is an equally weightedportfolio of volatility arbitrage funds. This is part of ourongoing research at Fimat where we study the returns offund managers running volatility arbitrage funds. TheFVAM was created as a performance benchmark for thehedge fund market - it is not an active fund as such.

TA: What funds make up the FVAM?

RH: There are seven volatility arbitrage funds within theFVAM: er Global, Lynx, Quadix, SGAM, Shooter Multi-Strategy, Titan Global and Turtle. These hedge funds treatvolatility as an asset class of its own.

TA: How widespread are volatility arbitrage funds withinthe hedge fund industry?

RH: Over the last few years we are seeing more and more.The first volatility arbitrage funds emerged in Europe earlyin 2002 and slightly earlier than that in the US. Fimat nowknow of about 20 volatility arbitrage managers around theworld.

TA: Does this mean global funds under management havereached significant levels?

RH: It is difficult to say because volatility arbitrage is trad-ed in many multi-strategy funds but as for pure volatilityarbitrage managers, it is probably somewhere in the regionof $3 billion globally.

VOLATILITY ARBITRAGE STRATEGIES

Rami Habib, quantitative analyst at Fimat in London, discussesvolatility arbitrage fund strategies and performance across the hedgefund industry.

RamiHabib

→→

Interview


TA: Why trade volatility? What are the return characteristicsof volatility arbitrage funds?

RH: Volatility itself has some attractive characteristics:

It is independent of the direction of the underlying pricemove.It is uncorrelated to the underlying securities.It increases when uncertainty increasesIt is mean revertingWhen most asset classes go down, volatility tends to goup

Volatility based strategies can offer uncorrelated risk-returnprofiles that most alternative investment and long-onlyportfolios require. Returns are also created without thecredit risk element that is inherent in conventional tradingsuch as convertible arbitrage.

TA: What are the typical strategies traded by volatility arbi-trage funds?

RH: There are many volatility trades that arbitrage fundscan enter into. Each fund manager will use his own tech-niques to extract value from movements in volatility, typical-ly by trading short-term, liquid and mostly exchange tradedfinancial securities and derivatives.

Specific volatility trading strategies include directionaltrading where managers can take directional views onvolatility, i.e. take an outright view (long or short) on thelevel of volatility (strikethrough). The trader may believethat options market expectations are wrong or thatimbalances in supply and demand for options have driv-en implied volatilities to levels where they no longerreflect consensus expectations of future realised volatili-ty. Some traders can switch from net long to net shortvolatility positions depending on their views on the mar-kets. Others will maintain a net long or short bias believ-ing the market systematically misprices tail risk.

Buying options in general means buying volatility. Deltahedging should be profitable if realized volatility minustransaction costs is higher than the implied level at whichthe options were purchased. Selling options in generalmeans selling volatility. Delta hedging should be profitableif realized volatility minus transaction costs is lower thanthe implied level at which the options were sold.

Relative value volatility trading involves taking relativevalue positions on volatility. The trader may take a viewthat the options on one stock or index are mispriced rel-ative to another. With volatilities remaining low, the rela-

tive value trades have been popular. In these markets,many traders are uncomfortable with selling volatility asthe profits are small but the risks are large. Some tradersare also avoiding buying volatility as it has remained lowover this year.

Term structure is where the trader trades on the shapeof the implied volatility term structures.

Skew trading involves a strategy of trading the relativelevels of implied volatilities across strikes - across theimplied volatility skew structure.

Dispersion/correlation trades. Dispersion trading allowstraders to profit from price differences using indexoptions and offsetting options on individual stocks.Dispersion trading can be explained by the fact that his-torically index volatility has traded rich, while individualstock volatility has been fairly priced. The dispersionstrategy typically consists of short selling options on astock index whilst simultaneously buying options on thecomponent stocks. Timing is key to a successful disper-sion trade. It is also important to carefully select the bas-ket of stocks for the offsetting dispersion basket. At thesimplest level they should account for a large part of theindex to keep the net risk low, but at the same time it iscrucial for the trader to make sure that they are buyingcheap volatility.

Finally, a new area that has only recently been looked atis cross-market volatility trades. Historically, volatilitytraders have been heavily focused on equities but we arenow seeing hedge funds looking at other asset classes forvolatility trading opportunities. For example, traders arelooking at the correlation between currencies and equi-ties or at taking volatility positions in the commoditymarkets. While, there won't be volatility in all the mar-kets all the time but there should be volatility is some ofthe markets some of the time.

TA: How have the strategies traded by fund managerschanged over the last few years?

RH: In the late 1980s and early 1990s the focus was ontheta and directional macro strategies using options insteadof stop losses. By the mid-1990s this had shifted to classicalvolatility arbitrage with delta hedging. The late 1990s andearly 2000 saw volatility arbitrage strategies trading acrossthe term structure, between plain vanilla and exotics, usingvolatility and variance swaps and more recently we are see-ing cross market volatility trading.

TA: What are the risks associated with volatility arbitrage?

•

••••

•

•

•

•

••

Interview


RH: One of the risks associated with vol trading is theimpact of changing volatility, this can affect the perform-ance of delta hedging. This is why it is important for tradersto monitor the vega as well as delta and gamma. The vegaof an option is the rate of change of the price of theoption with respect to the volatility of the underlying asset.For example, if everything remains constant and the mar-ket's view suddenly changes that future volatility is going tofall, then the option price will sit on a lower curve. This canbe good or bad news for the volatility player. Whatever thesituation, the portfolio will suffer an immediate mark tomarket loss due to the option price fall. However, if theunderlying price is near the exercise price then the increasein gamma can actually mean that there is more scope for re-hedging than before.

TA: How have implied volatility levels changed in the equitymarkets over the past decade or so? Have recent volatilitylevels been unusually low?

RH: Current equity market implied volatility is sittingaround 11-13%. By comparison to recent memory, this ishistorically low. but if one looks back further the currentlevels are above 'the 91 - 97' levels.

If one goes back to 1990 during the global recession,volatility for the S&P 500 index ranged from 20%-35%.During this period interest rates were high and low profitsmeant that debt coverage was low and corporate leveragewas high. Added to this is the lead up to the first Gulf Warwhich resulted in implied volatilities being highly priced.

Between 1991 and 1997 there was a market recovery.Interest rates were decreasing and this period was unevent-ful. As such implied volatilities spent a long period of timeat low levels, around 10-15%. From 1997 to 2003 therewere several events that resulted in spikes and high levels ofimplied volatilities. 1997 saw LTCM's collapse where the

closing of positions caused a liquidity crisis. In 1998 we hadthe Russian debt default and the Asian crisis which pro-duced implied vol levels of 50%.

From 1999 to 2001 there was the internet boom whenimplied volatilities ranged between 20-30% and 9/11 saw aspike at 45%. After that the Worldcom and Enron account-ing scandals and the Iraq War saw vols exceed 50% again.From 2003 to 2004 they declined rapidly as a result of lowactivity in the markets and a lack of major events.Moreover, interest rates were falling and there was renewedcorporate balance sheet strength.

TA: Do you see returns and opportunities in volatility arbi-trage diminishing because of the increased number of man-agers?

RH: Like any strategy with an arbitrage component thereturns may diminish as the number of factors increase. Acrowded trade remains a crowded trade. However, there isample capacity for asset class, instrument and geographicaldiversification in the years to come.

TA: What have the recent returns been like for volatilityarbitrage funds?

RH: Since the summer of 2002, volatility in the equity mar-kets has been in decline. As a result of the declining impliedvolatility levels and the fact that implied volatilities weretrading above realized market volatility levels, many of thelong biased volatility hedge funds experienced some losses.Some net short volatility funds managed to generate consis-tent returns over this period, however, care needs to betaken as volatility has the tendency to spike quite rapidly.Losses can be experienced as a result of prices gapping andthe market becoming illiquid at exactly the wrong time foroption sellers.

TA: What is your volatility outlook for these markets?

RH: Volatility is thought to be mean reverting. It is heavilydependent on consumer demand, interest rates and the levelof corporate investment leverage. As such, volatility couldeasily remain below 20% for quite a while. Consequently,the strategies that are expected to perform in this environ-ment are the long/short relative value volatility traders, themacro volatility traders and the dispersion traders in someof the non-arbitraged markets.

Rami Habib is quantitative analyst within AlternativeInvestment Solutions at Fimat International BanqueSA (UK Branch), in London.

“PURE VOLATILITY ARBITRAGE MANAGERS

PROBABLY HAVE SOMEWHERE IN THE

REGION OF $2 BILLIONUNDER MANAGEMENT

GLOBALLY.”- RAMI HABIB,

FIMAT INTERNATIONAL

In contrast, studies of the behav-iour of developing country realexchange rates are scarce. Thefew studies that have examinedthe determinants of these rateshave focused largely on LatinAmerica and have emphasizedthe role of terms of trade move-ments in driving the real exchange rate.

However, a natural assumption fordeveloping countries is that fluctua-tions in real commodity prices have thepotential to explain a large share ofchanges in real exchange rates, giventhat so many of these countries arehighly dependent on commodities —in some cases, a single commodity —for the bulk of their export revenues.Indeed, several studies have explored

this relationship for a handful of com-modity-exporting industrial countries,such as Australia, Canada, and NewZealand. But the biggest hurdle inextending these studies to developingcountries has been the lack of country-specific data on commodity exportprices.

That is why we undertook a study ofthe relationship between the realexchange rate and real commodityprices for all commodity-dependenteconomies. We asked the question: Doreal commodity prices and realexchange rates move together?

Identifying commodity currenciesOur study was based on the construc-tion of new monthly indices of nation-al real commodity export prices and thegathering of monthly real exchangerate data for 58 commodity-exportingcountries for the period January 1980to March 2002. Each country's (nomi-nal) commodity export price index is ageometric weighted average of world

prices for 44 individual nonfuel com-modities using country-specific exportshares (averaged over 1990-99) asweights.

The 58 commodity-exporting coun-tries include 53 developing countriesand 5 industrial countries, all of whichrely on commodity exports for a majorshare of their export income. Indeed,during the 1990s, the cross-countrymean share of total export receiptsderived from primary commodityexports was about 48 percent.Commodity exports typically exceeded50 percent of the total exports of sev-

eral sub-Saharan African coun-tries, especially Burundi (97 per-cent), Madagascar (90 percent),and Zambia (88 percent).

The share of primary commod-ity exports in total exports wasquite high even for the industrialcountries (Australia, 54 percent;Iceland, 56 percent). In addition,many countries remain over-whelmingly dependent on exportreceipts from their dominantexportable commodity — thedominant exportable exceeded 90percent of commodity exportreceipts in Dominica (bananas),Ethiopia (coffee), Mauritius(sugar), Niger (uranium), andZambia (copper).

Armed with the real exchange rateand real commodity export prices foreach country, we then checked to seewhether these two series displayed aclose relationship. We found that, formany countries, such as Australia andBurundi (both of which have flexiblenominal exchange rates), this wasindeed the case (see Figure 1), whileothers appear to display a relationshiponce a onetime movement in the real

For decades, economistshave tried with little suc-cess to model long-run

movements in real - that is,adjusted for inflation -exchange rates. Almost all ofthe studies have focused onindustrial countries, trying topinpoint whether fundamentalssuch as government spending,current account imbalances,and differences in productivityand interest rates hold the keyto explaining exchange ratemovements. But the resultshave been disappointing,with many models that arebased on fundamentalsfailing to provide a con-vincing explanation of thebehavior of real exchangerates in industrial countries.

Subject Matters


IDENTIFYING COMMODITY CURRENCIESby Paul Cashin, Luis Céspedes, and Ratna Sahay

Figure 1.

Which currencies can we accurately call "commodity currencies?"


Subject Matters

exchange rate (such as the 1994 devalu-ation for the CFA franc zone countriesof Mali and Togo) is accounted for.

Next, we used regression analysis toformally examine whether there was astable, long-run relationship between acountry's real exchange rate and the realprice of its commodity exports - inother words, which of the commodity-exporting countries qualified as having"commodity currencies"? In the analy-sis, we allowed for a structural shift inthe relationship between the two seriesto account for onetime movements ineither series (which typically involverapid movements in the nominal andreal exchange rate). We found just sucha long-run relationship for 22 of the 58commodity-exporting countries (seeTable 1).

It is not unexpected that sub-SaharanAfrican countries, given their depend-ence on commodity exports, accountfor half the commodity-currency coun-tries. Moreover, for these 22 countries,over 80 percent of the variation in the

real exchange rate can, on average, beaccounted for by movements in realcommodity prices alone — a surpris-ingly strong result. As for those com-modity-exporting countries where sucha long-run relationship could not befound, it is likely that factors other thanreal commodity prices played a key rolein real exchange rate movements.

How large an impact do real com-modity price movements have on thereal exchange rates of commodity-cur-rency countries? We found that theelasticity typically ranged between 0.2and 0.4, with a median of 0.38. Thus, a10 percent drop in the real price of thecommodity exports of countries withcommodity currencies is typically asso-ciated with a 3.8 percent depreciationof their real exchange rate. Furtheranalysis also indicated that, when devi-ations from the relationship betweenexchange rates and commodity pricesoccurred in countries with commoditycurrencies, they were caused primarilyby changes in real commodity prices.

Following a movement in commodityprices, it is typically the real exchangerate that then adjusts to restore its long-run relationship with real commodityprices.

A handy crystal ball Our study found evidence in supportof the co-movement of national realexchange rates and real commodityprices in a group of commodity-exporting countries. For these com-modity-currency countries, the worldprice of their commodity exports has astable and important effect on their realexchange rate. This empirical regularityis surprisingly robust, given the repeat-ed failure of previous attempts to usemodels that are based on fundamentalsto explain exchange rate movements.

For policymakers in commodity-exporting developing countries, under-standing the effects of commodityprice movements on exchange ratesshould be of great interest in guidingthe conduct of monetary and exchangerate policies, particularly as such coun-tries liberalize their capital markets andincrease the flexibility of theirexchange rate regimes. For countrieswith commodity currencies, commodi-ty prices are the key determinant oftheir real exchange rate and can be usedas a benchmark in determining whenexchange rates have deviated excessive-ly from their equilibrium value.

Paul Cashin is a Senior Economistand Ratna Sahay is an AssistantDirector in the IMF's ResearchDepartment. Luis Céspedes is anEconomist in the IMF's European IDepartment.

Extracted with permission from Cashin, P., Céspedes,L. and Sahay R., (2003), "Commodity Currencies",Finance and Development, Vol 40, No. 1, IMF.

Table 1. Identifying the commodity-currency countries. Countries found to have commoditycurrencies are in bold. Two-fifths of commodity-exporting countries have commodity curren-cies: there is a long-run relationship between their real effective exchange rate and their real com-modity export prices. Source: Cashin, Céspedes, and Sahay, 2002.

BurundiCameroonCentral African RepublicCôte d'IvoireEthiopiaGhanaKenyaMadagascarMalawiMaliMauritaniaMauritiusMozambiqueNigerSenegalSouth AfricaSudanTanzaniaTogoUgandaZambiaZimbabwe

AustraliaBangladeshIndiaIndonesiaMalaysiaMyanmarNew ZealandPakistanPapua New GuineaPhilippinesSri LankaThailand

MoroccoSyrian Arab RepublicTunisiaTurkey

IcelandNorway

ArgentinaBoliviaBrazilCanadaChileColombiaCosta RicaDominicaEcuadorGuatemalaHondurasMexicoNicaraguaParaguayPeruSt. Vincent and GrenadinesSurinameUruguay

Sub-SaharanAfrica

Asia-Pacific

Middle East andNorth Africa

WesternHemisphere

Europe

Software


Dresdner Kleinwort Wasserstein's

Banks and financial insti-tutions are increasinglylooking to leverage

their in-house market researchin the most effective way possi-ble. The growth of the hedgefund industry and competitionamongst prime brokerage serv-ices means that greater value isnow attached to research out-put. This has meant that moreinstitutions are now developingtheir own proprietary researchtools in order to give their

front- end services an edge overthe competition.

At Dresdner Kleinwort Wassersteinin London, Max Knudsen, head ofcapital markets technical strategy, hasdeveloped PIA - Price InformationAdvantage. PIA is DrKW's propri-etary interactive sentiment forecast-ing service designed to optimise thetiming of trade execution. The serv-ice provides technically derived tim-ing and trading recommendations viaa daily interactive email service. PIAis based on an assessment of marketsentiment in the equity, FX and inter-est rate markets made by Knudsenand his three man team. Sentiment is

assessed using a combination of tra-ditional charting techniques includingBollinger Bands, momentum and sto-chastics.

Knudsen says the role of his teamis to translate their assessment ofmarket sentiment (majority bullish-ness and bearishness) in each marketand to convert this into specific trad-ing and timing advice consisting of aforecast for the direction of senti-ment and the most important prices."This is 75% a statistical and technicalprocess" says Knudsen, "with theremaining 25% drawn from practicalmarket experience. Each member ofmy team has around 20 years

PRICE INFORMATION ADVANTAGE

PIA Team, from left: Pavel Gronbjerg, Max Knudsen seated, Steve Lucas, Alan Collins

→→


Software

market experience in either sales ortrading and I consider this crucial inbeing able to accurately assess marketsentiment."

PIA was developed in response tocomments from clients that theyreceived too much research, much ofwhich looked indistinguishable.Simplicity seems to be the key toPIA. There is very little text; juststraight forward recommendations ofwhen to buy or sell and the 3 mostimportant support and resistance lev-

els. "When we started developingPIA 5 years ago" says Knudsen, "itbecame clear that clients felt theywere already receiving too muchresearch. Our goal was to design aresearch application that was bothreliable and very easy to interpret anduse".

PIA covers 6 major currency rates,6 stock indices and 9 interest ratemarkets (see Table 1.) DrKW mayextend the services to cover the cred-it and commodities markets some-

time next year."By looking at price action and

price development we are able toidentify how sentiment is changing"adds Knudsen. "We discern what cur-rent sentiment is from a technical andstatistical analysis of price actionalone. This provides us with 75% ofthe information we need. Ourassumption of the impact differingprice changes will have on currentmarket sentiment makes up theremaining 25%. This allows us topublish an implied confidence levelwith each recommendation." Theconfidence levels range from 65% to80% and reflect Knudsen's and histeam's assessment of the reliability ofeach call. However as Knudsenstresses, "This service is all aboutmarket timing, we don't pretend to befortune tellers. The service isdesigned to interpret, track and spotchanges in majority market sentimentand to assess the likely impact on thedirection and timing of investoractivity".

PIA now has 1300 registered users.Among them is Nick Gartside, aEuropean government bond fundmanager with Schroders in London."We get an awful lot of technicalresearch from various sources" saysGartside, "although with each recom-mendation that's published, thedegree of bullishness of bearishnessis seldom mentioned. PIA publishesconfidence percentages with its levelsand this helps compliment my owntechnical research”.

Using candlesticks Time periods covered by PIA areintraday, weekly and quarterly. Thissuits the DrKW client base served byKnudsen's team. Their clients rangefrom bank proprietary desks andhedge funds to real money accounts,corporates and central banks. Eachmarket and time period features up to3 candlestick charts containing a min-imum of conventional chart analysissuch as the plotting of trendlines and

Fixed income/MM FX Equities

10-yr Treasury EUR/GBP FTSE BOBL EUR/JPY DAX

Bund EUR/USD S&P 500

Euribor GBP/USD Eurostoxx Eurodollar USD/CHF Nasdaq

Schatz USD/JPY Nikkei

Short Sterling

T Bond

Eurodollars

Table 1. PIA market coverage

Figure 1. The PIA fixed income page for weekly Bunds

Software


pattern identification. This allowsprice action to be clearly highlightedwithout the clutter often associatedwith price charts, although non-can-dlestick aficionados may still yearnfor conventional bar chart analysis.However, Knudsen is confident thatcandlesticks provide the most practi-cal charting method. "With candle-sticks we are able to visually differen-tiate the positive and negative periodsplus highs and lows by using contrast-ing colours. Clients tell us this makesour research that much easier to dis-seminate than our competitorsresearch" he says. "We cannot achievethis with bar charts. Furthermore, theopen and closing price of a markethave an important relationship tomarket sentiment. They allow us to

identify and track this relationshipmore closely than would bar charts."

Figure 1. is a screenshot for a PIAreport on weekly Bunds showing thetrading recommendation (sell below123.78) with its confidence level(70%). Figure 2. is a PIA 'Cheat Sheet'which provides an at-a-glance sum-mary of all markets covered by theservice and its basic recommenda-tions.

Not real-time PIA is not a real-time service. Eachreport is prepared the night beforethe market opens and is publishedand delivered by email that night.Although the US FX market is stillopen by the time PIA reports havebeen written, the trading recommen-

dations are still valid for the next day'strading. As Knudsen explains, "overthe past 4 years trading in the US andAsian markets rarely has a detrimentalimpact upon the calls we have madefor the following day in Europe. FXactivity is really centred in Europe soit would take a significant event in theUS or Asia for us to alter our fore-casts the following morning.However when this does occur we re-analyse sentiment and send out anupdated forecast as quickly as possi-ble, usually by 8am London time.

Knudsen points out that PIA hasnot been designed as a tip sheet. "Ihave never considered this as my role- that is trading." He concludes "Mygoal from the outset was to offerclients a consistently reliable call onjust two things: the direction and tim-ing of sentiment changes. The resultsof our efforts in the 4 years since westarted are a 73% hit rate from a totalof 11,908 calls. With these results wecan make a valuable contribution to aclient's decision of when and what totrade."

For more information contact:[email protected]

Figure 2. The PIA Cheat Sheet

“PIA PUBLISHES

CONFIDENCE

PERCENTAGES WITH

ITS LEVELS AND THIS

HELPS COMPLIMENT

MY OWN TECHNICAL

RESEARCH.”

NICK GARTSIDE, SCHRODERS

21/05/01 –12/07/05

Fixed income FX Money markets Equities

No. of calls 3742 6042 799 1325

No. correct 2638 4498 573 955

Hit rate 70% 74% 72% 73%

Table 2. PIA hit rate

The CubeThe browser-based

Market Data Solution.

Real-time Data, News, Analytics, charts and the TraderMade database available globally 24/7, fully customisable by you.

Contact us at:

tel: +44 (0)20 8313 0992

email: [email protected]

Book Review


Mechanical Trading SystemsPairing trader psychology withtechnical analysis

By Richard L WeissmanJohn Wiley & Sons, Inc.217 pages, £60.00ISBN 0-471-65435-3

Mechanical Trading Systems isavailable from the TechnicalAnalyst Bookshop at the reducedprice of £51.00 plus £2.00 P+P. Toorder please call 01730 233870and quote "The Technical Analystmagazine" or order online atwww.global-investor.com/techni-calanalyst. This book is usuallyposted within three working daysof your order.

When I picked this book up and leafed through its content it rang all the alarm bells Iuse to detect groan-inducing trading books. Each chapter starts with a quote from thelikes of Aesop, Sun Tzu, and those great giants of trading, Shakespeare and Buddha.There is a summary of technical analysis in chapters 1 to 3 that is essentially a re-hash ofexisting knowledge (although I understand the need), words like "utilization" are alwayspreferred to their simple alternative such as "use", and there is a promise of "pairing trad-ing psychology with technical analysis". It all looked very familiar. So I certainly wasn'texpecting it to be the outstanding book that it is.

Weissman's book does what it says it's going to do in the title. That is, it tells you howto use mathematically-based technical analysis such as moving averages, RSI andBollinger Bands to create a profitable mechanical trading system.

The trading systems are divided into two types - trend following strategies and meanreversion strategies - across four time scales (long-term, intermediate, swing, intraday).Helpfully - for users of CQG at least - it also includes simple programming language forCQG backtesting and optimisation software, although I wouldn't let this put you off ifyou use another charting provider.

This division into trend-following and mean-reversion strategies is not simply an organ-isational device. With each category, he talks about the mentality needed to stick with astrategy, making reference not just to profit and risk/reward but to all the statistics thatreally help understand what it would be like to trade the strategy - number of consecu-tive losing trades, maximum drawdown amount, time out of the market, percent winners,average trade duration, etc. This may sound dry, but with Weissman's authoritative styleand understanding of what it means to be a trader, he makes the statistics come alive andyou can see the trader enduring the awful losses one after another. This is no gimmick -he really has paired trading psychology with technical analysis and presented it in such away that you can properly assess what trading style is best suited to you.

Furthermore, Weismann's clarity makes this one of the most thought provoking andenjoyable books on trading. Its advice is practical and simple to apply and it will spur youon to ask deeper questions (regarding trading at least) that will probably lead to experi-mentation and success.

One question that emerges is: are the strategies outlined in this book just ways of mak-ing the trader apply sensible risk management discipline, rather than any great tool toexploit market anomalies? The winning ratios, particularly for trend-following tech-niques, are very low and suggest poor forecasting ability, despite their profitability. Manytraders might shrug their shoulders and say that technical analysis was never about mak-ing forecasts.

However, other techniques such as chart patterns and Elliott Waves and even trend-lines, do not feature in this book, (nor in Lars Kestner's equivalent book "QuantiativeTrading Strategies" reviewed in the last issue). Yet these methods are still mathematical,albeit they are difficult to model. How much more profitable could strategies be if theywere able to draw on this other non-linear side of technical analysis that arguably offersmore predictive capability?

This book is highly recommended for almost anyone involved in trading, forecasting,or studying the markets. For the trader in particular, reading this book should help iden-tify and emphasize the psychological traits that will be needed to apply a coherent strat-egy. It will also provide a framework for assessing strategies, and cfor reating new andbetter ones.

MECHANICAL TRADING SYSTEMS


Commitments of Traders Report

COMMITMENTS OF TRADERS REPORT6 July 2004 - 12 July 2005Futures only (open interest) non-commercial net long positions and spot rates

10-year US Treasury Source: CBOT

Dow Jones Industrial Average Source: CBOT

5-year US Treasury Source: CBOT

Swiss franc Source: CME

Pound sterling Source: CME Yen Source: CME

-250000

-200000

-150000

-100000

-50000

0

50000

100000

150000

200000

06/07/2004 28/09/2004 21/12/2004 15/03/2005 07/06/2005

3.40

3.60

3.80

4.00

4.20

4.40

4.60

4.80

Non-commercial net long

Spot

-200000

-150000

-100000

-50000

0

50000

100000

150000

200000

250000

300000

06/07/2004 28/09/2004 21/12/2004 15/03/2005 07/06/2005

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5Non-commercial net long

Spot

-10000

-5000

0

5000

10000

15000

20000

06/07/2004 28/09/2004 21/12/2004 15/03/2005 07/06/2005

9800

10000

10200

10400

10600

10800

11000


Spot

-50000

-40000

-30000

-20000

-10000

0

10000

20000

30000

40000

50000

06/07/2004 12/10/2004 04/01/2005 29/03/2005 21/06/2005

1.00

1.05

1.10

1.15

1.20

1.25

1.30

1.35Non-commercial net long

Spot

-30000

-20000

-10000

0

10000

20000

30000

40000

50000

06/07/2004 28/09/2004 21/12/2004 15/03/2005 07/06/2005

1.65

1.70

1.75

1.80

1.85

1.90

1.95

2.00Non-commercial net long

Spot

-80000

-60000

-40000

-20000

0

20000

40000

60000

06/07/2004 28/09/2004 21/12/2004 15/03/2005 07/06/2005

96

98

100

102

104

106

108

110

112

114


Spot


Commitments of Traders Report

Euro Source: CME

Nasdaq Source: CME

3-month eurodollar Source: CME

Nikkei Source: CME

Gold Source: CEI US dollar index Source: NYCE

-20000

-15000

-10000

-5000

0

5000

10000

15000

20000

25000

06/07/2004 28/09/2004 21/12/2004 15/03/2005 07/06/2005

104

106

108

110

112

114

116

118


Spot

0

20000

40000

60000

80000

100000

120000

140000

160000

06/07/2004 28/09/2004 21/12/2004 15/03/2005 07/06/2005

300

320

340

360

380

400

420

440

460

480

500


Spot

-4000

-2000

0

2000

4000

6000

8000

10000

06/07/2004 28/09/2004 21/12/2004 15/03/2005 07/06/2005

10000

10500

11000

11500

12000

12500


Spot

-800000

-600000

-400000

-200000

0

200000

400000

600000

06/07/2004 28/09/2004 21/12/2004 15/03/2005 07/06/2005

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

4.00


Spot

-10000

-5000

0

5000

10000

15000

20000

06/07/2004 28/09/2004 21/12/2004 15/03/2005 07/06/2005

9800

10000

10200

10400

10600

10800

11000


Spot

-30000

-20000

-10000

0

10000

20000

30000

40000

50000

60000

70000

06/07/2004 28/09/2004 21/12/2004 15/03/2005 07/06/2005

1.10

1.15

1.20

1.25

1.30

1.35

1.40


Spot

Long-Term Technicals


LONG-TERM TECHNICALS

Provided by Thomas Anthonj, ABN Amro, Amsterdam

Stalling right under the projected target zone for a potential top(1.9566-88), the market retreated and is currently sitting down onkey-support at 1.7308. This should provide support in should pricesbegin to rally. Breaking below would only leave minor support at1.7025 and 1.6904 (Fib.support/old top) to prevent a much biggersetback towards 1.5923 and 1.5066 (61.8/76.4 %). So above1.7308 we still see a good chance for another advance towards2.0115 and 2.0165 (old top/Fib.projections) which should definitelycap the market for a while.

GBP-USD

Forming almost a double bottom close to the historic 101.25 bottomand breaking decisively above trend line resistance two months ago,the market already indicated a bigger rebound. But whether the up-momentum is strong enough to clear key-resistance at 114.46 is tooearly to say. A break above would reverse the earlier 3 year longbear-trend whereas a failure to do so would most likely lead toanother sell-off targeting 101.25 and the head-and-shoulders targetat 95.75. .

USD-JPY

Retreating from a projected Fibonacci-target for a major top (1.3650)the market entered a bigger consolidation pattern that has alreadycome very close to the decisive support zone between 1.1760 and1.1590. Above this we expect a minimum rebound up to 1.2930 topotentially form the right shoulder of a bigger head-and-shoulderspattern. Only above 1.2930 would we see evidence that the old up-trend has resumed. However, a break below 1.1760/1.1590 wouldcall for a much bigger setback towards 1.1000/1.0764 and1.0380/00.

EUR-USD

The triangle consolidation observed lately is still leaving the back-door open for a continuation of the up-trend (target 490) what wouldbe confirmed on a triangle breakout at 441. In case triangle supportat 416 should fail, we'd have to expect a minimum setback towards379 before the bulls might get another chance.

Gold


Long-Term Technicals

Having spent more than a year in a sideways consolidation pattern,the market managed to break and close above long-term trend lineresistance (now support at 11509). This supports the idea that itwants to break out of an inverted head-and-shoulders reversal pat-tern that would project a target at 16850 once the neck/trend line isbroken on close at 12307. There under we are still running a risk ofmissing the right shoulder of the head-and-shoulders pattern thatwould imply another decline to around 9383.

Nikkei

Retreating from a potential target zone for the first bigger bull-cycleup between 11036 and 11400, the market is still in a vulnerablestage as the correction of this potential accumulation phase mighthave just started. However, as long as the market manages to stayabove trend line support at 10165, the bulls are still in controlshooting for the next two major tops at 11350 and 11750 whichagain are big hurdles to overcome. A break and close below trendline support would be a cause for concern as we might be due fora 61.8 % or 76.4 % retracement of the whole advance from 7197.Having said that we'd still have room for a decline to 8644 andpotentially even 8091/62.

Dow Jones

The setback experienced from 2192 down to 1890 in the first 4months of this year is much too small to be classified as a correc-tion based on the whole advance from 1108. The conclusion istherefore twofold: A) The market is shortly going to break abovethe last 2192 top which should trigger further buying, and if the old2328 top can also be cleared, strong acceleration upwards. B) Butif 2192 can't be cleared we are still running a fairly high risk ofaccelerating down with a minimum target at 1778/51.

Nasdaq

The latest break above the March top at 1229 is very positive for thecurrent chart. But still trading within a Fibonacci-target cluster for apotential top of the bull-cycle from 768, we are on alert for signs ofweakness such as a trend line break at 1179. Such a break couldsignal the end of this accumulation phase. If the last low at 1136 isalso taken out we'd have to expect a minimum setback to 1061 andpotentially even to 954/45 (old top/61.8 %/old low). In order to delaythe permanent risk of a bigger setback, the market would have toclear strong resistance between 1246 and 1288(Fib.projection/retracement/trend line).

S&P 500


Technical Analysis in the

Commodity & FX Markets

August

23

Cape Town,

South Africa

[email protected]

+44 (0) 207 833 1441



October

25

London,

UK

[email protected]

+44 (0) 207 833 1441

Technical

AnalysisAugust

4-5

New York,

[email protected]

Monthly

Meeting

October

12

London,

UK

[email protected]

Annual

Dinner

September

22London,

UK

[email protected]

Monthly

Meeting

September

14

London,

UK

[email protected]


Commodity & FX Markets August

25

Johannesburg,

South Africa

[email protected]

+44 (0) 207 833 1441



September

27Dubai,

United Arab

Emirates

[email protected]

+44 (0) 207 833 1441

Organiser Date Event Venue Contact

EVENTS 2005

Events

drkw’s technical strategistsusage increases sharply for larger firms with $70 billion and above...

Documents