sebi proposals and how they might affect liquidity providing market makers 20160830
TRANSCRIPT
Why the new regulations proposed for Algorithmic Trading and Co-Location by SEBI
effect Liquidity Providing Market-Makers
… and would therefore dramatically alter the entire market place.
Rajib Ranjan Borah, Partner – iRage Broking Services LLP
Structure of this presentation
Phase 2
• Impact of new proposed SEBI regulations on market making, liquidity and overall market.
• Pros and cons of each proposal – likely beneficiaries and losers
• Case study of how manipulators could misuse additional complexity introduced in each of the new proposals
Phase 1 • Introduction – types of market participants • The role of market makers. • How algorithmization of market making has improved the
overall market eco-system
Types of Market Participants
Let us first look at the different types of financial market participants:
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
Banks, insurance companies, retirement funds, hedge funds,
mutual funds, etc that are financially sophisticated & make
large investments Investor (institutional)
Individual investor who trades for his/her own personal
account Investor (retail)
Buyer / seller of financial instruments who is aiming to profit
from fluctuations in prices by making guesses (instead of
owning assets). Trader (speculator)
Participants reducing their exposures by trading in financial
markets (e.g. importers/ exporters, consumers of
commodities as raw materials, investors, etc) Hedger
Participants who provide buy & sell quotes in financial
instruments for counterparties – hoping to earn the difference
between the sell and buy quote prices Market Maker
Participants who aim to profit by trading against incorrect
prices in the market (and hoping to profit when prices rectify)
– thereby bringing pricing efficiency Trader (arbitrageur)
Types of Market Participants doing Algorithmic Trading
Amongst these, the predominant users of algorithmic trading are the following (algorithmic trading: use of computer algorithms to execute their trades)
Banks, insurance companies, retirement funds, hedge funds,
mutual funds, etc that are financially sophisticated & make
large investments Investor (institutional)
Individual investor who trades for his/her own personal
account Investor (retail)
Buyer / seller of financial instruments who is aiming to profit
from fluctuations in prices by making guesses (instead of
owning assets). Trader (speculator)
Participants reducing their exposures by trading in financial
markets (e.g. importers/ exporters, consumers of
commodities as raw materials, investors, etc) Hedger
Participants who provide buy & sell quotes in financial
instruments for counterparties – hoping to earn the difference
between the sell and buy quote prices Market Maker
Participants who aim to profit by trading against incorrect
prices in the market (and hoping to profit when prices rectify)
– thereby bringing pricing efficiency Trader (arbitrageur)
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
What is HFT
Within Algorithmic Trading, a special category of trading where the need for information (and the need to respond to information) is quick, is known as High Frequency Trading (HFT)
Investor (institutional)
Investor (retail)
Trader (speculator)
Hedger
Market Maker
Trader (arbitrageur)
HFT Algorithmic
Trading
Typical Characteristics of HFT: • Quick response to market events • Small trading positions / positional bets
• High Turnover – i.e. doing a lot of round trips (both buy & sell trades)
• Huge investments in technology
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
Different types of HFT strategies
Each of the 5 types of participants that use algorithmic trading also do HFT Regulations on HFT (and algorithmic trading) will therefore impact all the five groups
Execution HFT
Strategies Investor (institutional)
Investor (retail)
Order flow
dynamics HFT
Execution HFT
Strategies Hedger
Automated HFT
Market Making Market Maker
Automated HFT
Arbitrage Trader (arbitrageur)
Trader (speculator)
• There are different types of
HFT strategies • The end objectives &
underlying philosophies of different types of HFT strategies are different
↓ • Applying the same
regulations on all types of HFT strategies will effect groups for which the regulations were initially not intended
Type of HFT strategies
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
Introduction to Market Making
Who are market makers
How do market makers earn profits?
What are the market makers’ risks?
How algorithmic trading enables market
making
Why we need algo market makers
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
Will buy @Rs 100
Will sell @Rs 110
Introduction to Market Making
Who are market makers
How do market makers earn profits?
What are the market makers’ risks?
How algorithmic trading enables market
making
Impact of algo market makers on markets
Market makers are entities that provide both buy & sell quotes for others
MM
Essentially, market makers take the opposite side of investor trading volume. For investors who want to buy, market makers will sell to them. For investors
who want to sell, market makers will buy from them.
Will buy @Rs 100
Will sell @Rs110
MM
Need to buy @ Rs 110
MM
OK … I sell @Rs 110 to you
OK… I buy @ Rs 110 from you
Market makers, therefore, satisfy the supply & demand of the financial markets & keep securities changing hands between sellers & buyers.
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
Introduction to Market Making
Who are market makers
How do market makers earn profits?
What are the market makers’ risks?
How algorithmic trading enables market
making
Impact of algo market makers on markets
MM
Market makers aim to buy at their ‘bid’ quote price & sell at their ‘ask’ quote price. The difference between the ‘bid’ and ‘ask’ prices is the ‘bid-ask spread’, and that is the profit that market makers target
Will buy @Rs 100
Will sell @Rs 110
If the market maker successfully (i) sells at Rs 110
(ii) & buys at Rs 100 Then he makes a profit of Rs 10 after those two trades
Ask
Bid
Bid – Ask Spread
In many developed economies, there are official market making schemes - wherein participants meeting certain quoting obligations are
provided financial rewards. India does not have market making schemes (with a few small
exceptions), but traders do execute market making strategies (despite absence of external financial rewards) - hoping to earn bid-ask profits.
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
Introduction to Market Making
Who are market makers
How do market makers earn profits?
What are the market makers’ risks? -1
How algorithmic trading enables market
making
Impact of algo market makers on markets
MM
Since market makers are always counter parties to trades done by informed traders
In case of price moves, the market makers are often stuck with wrong positions
Prices have gone up after I sold (to an informed trader),
I can only buy @ Rs 125 now !
MM
I have sold @Rs 110
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
Introduction to Market Making
Who are market makers
How do market makers earn profits?
What are the market makers’ risks? -2
How algorithmic trading enables market
making
Impact of algo market makers on markets
MM
The biggest risk for the market maker is not having the latest information
Will sell @Rs 110
Will buy @Rs 100
I will buy @ Rs 110 immediately
Ah… the MM is still willing to sell @ Rs 110
– prices should be higher after event X
For market makers to survive by managing risks, it should be possible for them to receive & respond to information quickly.
Introduction of any speed barrier will hinder this ability
For a strong market, it is necessary to have market makers.
To have market makers, it should be possible for them to survive & succeed without big losses
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
Introduction to Market Making
Who are market makers
How do market makers earn profits?
What are the market makers’ risks?
How algorithmic trading enables market
making - 1
Impact of algo market makers on markets
To be efficient, market makers should be able to - adjust their quotes immediately in response to events*
* Events could be (i) changes in prices of financial instruments, (ii) trading positions accumulated by the market
maker
Automated systems are more efficient than human beings in detecting & responding to such events.
Therefore they can handle their risks better
1. Since automated systems can handle their risks better, therefore they offer better* quotes for others
* better = closer to fair valuation
Will buy @Rs 104.5
Will sell @Rs 105.5
Bid – Ask Spread = Rs 1 Automated
MM
Will buy @Rs 100
Will sell @Rs 110
Bid – Ask Spread = Rs 10 MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
Introduction to Market Making
Who are market makers
How do market makers earn profits?
What are the market makers’ risks?
How algorithmic trading enables market
making - 2
Impact of algo market makers on markets
Pricing of derivatives that enable investors to hedge often involve time consuming mathematical calculations
While humans can take minutes, automated systems are can do these calculations in microseconds
MM
Automated MM
Give me more time to
calculate my quotes
My quotes are ready.
I will buy @Rs 10 & sell @ Rs
10.1
2. Response time is therefore much faster
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
Introduction to Market Making
Who are market makers
How do market makers earn profits?
What are the market makers’ risks?
How algorithmic trading enables market
making - 3
Impact of algo market makers on markets
Human traders can only track activities in a few instruments, while automated systems can do thousands simultaneously
The same trader using an automated trading system provide liquidity in significantly more financial instruments simultaneously
MM
Automated MM
Instrument A: I will
buy @A1 & sell @A2
Instrument B: I will
buy @B1 & sell @B2
Instrument C: I will buy @c1 & sell
@c2
Instrument D: I will buy @d1 & sell
@d2
Instrument F: I will buy @f1 & sell
@f2
Instrument G: I will buy @g1 & sell
@g2
Instrument A: I will buy @a1 & sell
@a2
Instrument B: I will buy @b1 & sell
@b2
Instrument E: I will buy @e1 & sell
@e2
Instrument H: I will buy @h1 & sell
@h2
3. Scalability is therefore much higher
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
Introduction to Market Making
Who are market makers
How do market makers earn profits?
What are the market makers’ risks?
How algorithmic trading enables market
making - 4
Impact of algo market makers on markets
Machines don’t take lunch & toilet breaks
Automated market making systems are always active
MM
Automated MM
Let me come back from lunch
Instrument C: I will buy @c1 & sell
@c2
Instrument D: I will buy @d1 & sell
@d2
Instrument F: I will buy @f1 & sell
@f2
Instrument G: I will buy @g1 & sell
@g2
Instrument A: I will buy @a1 & sell
@a2
Instrument B: I will buy @b1 & sell
@b2
Instrument E: I will buy @e1 & sell
@e2
Instrument H: I will buy @h1 & sell
@h2
4. Availability / uptime is therefore higher
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
Introduction to Market Making
Who are market makers
How do market makers earn profits?
What are the market makers’ risks?
Difference between prices of consecutive trades done against a human market maker will be much higher
than those done against an automated market maker
1. Asset price volatility therefore reduces
MM And then bought @Rs 100
Sold @Rs 110 Automated MM
And then bought @Rs 104.5
Sold @Rs 105.5
1. Price volatility = 110 to 100
2. Excess spread paid = Rs 5
1.Price volatility = 105.5 to 104.5
2. Excess spread paid = Rs0.5
How algorithmic trading enables market
making
Impact of algo market makers on markets -1
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
Introduction to Market Making
Who are market makers
How do market makers earn profits?
What are the market makers’ risks?
With automation rendering market making easy, order books have become thick.
Execution price for even big orders are close to fair price
Buy @Rs 104.5
Sell @Rs 105.5 Automated
MM 1
Buy @Rs 104.4
Sell @Rs 105.6
Automated MM 2
I am able to trade a huge size at little
impact cost
Buy @Rs 104.3
Sell @Rs 105.7
Automated MM 3
2. Impact cost & volatility is thus lower
How algorithmic trading enables market
making
Impact of algo market makers on markets -2
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
Introduction to Market Making
Who are market makers
How do market makers earn profits?
What are the market makers’ risks?
How algorithmic trading enables market
making - 4
Impact of algo market makers on markets -3
Made markets
less Volatile
Reduced indirect
costs paid as bid-ask spreads
Increased liquidity
across the board
Algorithmic Market Making has
therefore provided the following
benefits:
Introduced liquidity in
hedging derivatives
Efficiently Priced
Markets (by interaction
with arbitrageurs)
Reduced impact cost
(while trading big sizes)
Combining all the previous slides …
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals
Structure of this presentation
Phase 2
• Impact of new proposed SEBI regulations on market making, liquidity and overall market.
• Pros and cons of each proposal – likely beneficiaries and losers
• Case study of how manipulators could misuse additional complexity introduced in each of the new proposals
Phase 1 • Introduction – types of market participants • The role of market makers. • How algorithmization of market making has improved the
overall market eco-system
1. Minimum Resting Time for Orders
Impact on investors & market
Possibility of misuse & fraud
Overall assessment
About Proposal
Impact on market makers
Proposal: As per the Minimum Resting Time mechanism, the orders received by the stock exchange would not be allowed to be amended or cancelled
before a specified amount of time viz. 500 milliseconds is elapsed (also known as resting time)
Objective 1: This issue of ‘fleeting’ or ‘vanishing’ liquidity arises from the ability of the trading algorithms to react to new developments by modifying /
cancelling / placing new orders very quick
Slower participants therefore cannot use the liquidity provided by trading algorithms because the quote prices have changed / vanished
Objective 2: Ability to modify orders quickly has raised concerns with a section of market participants who consider that this ability is prone to market abuse.
Precedent: Australian Securities & Investment Commission (ASIC) sought feedback few years ago, but decided not to go ahead with the proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 1 of 7)
1. Minimum Resting Time for Orders
After the new proposal: When an event happens immediately after a market maker posted his quotes
The market maker will not be able to modify their quotes
till the resting period of his existing quotes is over
Liquidity taker Let me trade against the
market maker’s incorrect quotes
To protect themselves, market makers might either start quoting wider.
Some might quit !
My sell quote is @Rs 120 instead of Rs 105
My buy quote is @ Rs 90
instead of Rs 104
Automated MM
Currently, algo market makers have the confidence that they will be able to change their quotes quickly (in response to events).
Therefore they provide better quotes (lower bid-ask spread)
Will sell @Rs 105 Will buy @ Rs 104 Automated
MM
Event X (new fair price is 107)
Need to increase my quotes after event X to Rs 106 & Rs 108 Automated
MM
But I have to wait till resting period
These will be exploited by other market participants. The market maker will therefore be hit on the wrong
side
Automated MM
Sold @ Rs 105 for a loss
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 1 of 7)
All disadvantages associated with reduced market making are going to happen - high volatility, poor price discovery, high
impact cost, high spread cost
1. Minimum Resting Time for Orders
1
1
My sell quote is @Rs 120 instead of Rs 105
My buy quote is @ Rs 90
instead of Rs 104
Automated MM
What I could buy @ Rs 105 earlier, I have to buy @ Rs 120 now !
Current Bid – Ask Spread = Rs 30 Earlier Bid-Ask Spread = Rs 1
Impact on investors & market -1
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
However, market makers will respond by quoting wider & safer, therefore no one is going to benefit much
Traders who exploit forced delays on market makers will benefit (Often liquidity takers who send a lot of IOC orders hoping for misprices)
2
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 1 of 7)
1. Minimum Resting Time for Orders
Impact on investors & market -2
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Mayhem, increased volatility at the moment of events
3
Traders whose quotes are within resting period & rendered mispriced after a new event might get wrong positions
To nullify such positions, they would try to do trades in the other direction by sending new orders in the right direction
1. My sell quote is within resting period and prices are going up
Automated MM
3. Let me buy simultaneously & offset the trade forced on me
2. My sell quote is going to get hit & I am going to get a sold position
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 1 of 7)
1. Minimum Resting Time for Orders
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud
E.g.: The price of options instruments is linked to the price of the
future instrument. ↓
A market maker places orders in all options based on the prices in the future instrument.
↓ (Assume the future & options are not very liquid)
Immediately after the market maker’s orders, a manipulator will manipulate the price of the future instrument (say upward).
And also trade against all the market maker’s option quotes (which are still under resting phase)
↓ The market maker will have to hedge the delta from the option
trades by buying the futures ↓
The manipulator will sell the future and close the position
This is not possible currently, because the moment the prices in the future instrument changes, the market maker immediately
modifies the prices in the options
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 1 of 7)
1. Minimum Resting Time for Orders
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Beneficiaries
Losers
• Liquidity takers who exploit mispricing forced on other traders (by restrictions on quote updates)
• Manual traders – manual market makers who existed prior to automated market makers
• Foreign exchanges
• Market Makers • Retail Investors • Investment Funds • Indian exchanges because of reduced volumes
Summary
Overall assessment
• Problem of fleeting orders happens at the time of activity and is not linked to algorithmic trading. Event manual traders quit at the time of high activity Two independent researches done by Susan Thomas (IGIDR) and Prof Ashok Bannerjee (IIM Calcutta) show that during moments of activity in India non-algorithmic traders quit the market, and algorithmic traders stayed
• Any concern related to manipulation by changing quotes quickly should be monitored & prevented – rather than introducing hindrances that impact everyone (including genuine liquidity providers)
• Allegations about vanishing orders are not valid in India - these are possible in US, etc because of a combination of factors (NBBO, flashing, complex order types,
rebates, etc). Without detailed analysis, it is easy for Indian firms to feel that they would be ‘gamed’ similarly (fuelled by excessive reporting of events from the US)
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 1 of 7)
2. Frequent Batch Auctions
Proposal: Frequent Batch Auctions would accumulate buy & sell orders on the order book for a particular length of time (say 100 milliseconds).
At the end of every such period, the exchange would match orders received
during the time interval.
Objective: Set a time interval for matching of orders which is short enough to allow for opportunities for intraday price discovery, but long enough to
minimize the latency advantage of HFT to a large extent.
Impact on investors & market
Possibility of misuse & fraud
Overall assessment
Impact on market makers
About Proposal
Precedent: Taiwan Stock Exchange (TWSE) used to have continuous auction mechanism – however this has been discontinued
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 2 of 7)
2. Frequent Batch Auctions
Currently: To make markets in multiple derivatives
that derive their prices from a single underlying instrument: Market makers use the instantaneous prices of the underlying
to compute prices of all the derivatives.
When the prices in the underlying instrument change, market makers modify their quotes in all derivatives
After proposal: For the duration when the market is frozen because of the auction, market makers would not have the latest price of the underlying.
On each auction, the market maker is therefore making markets using stale prices of the underlying (i.e. prices from the previous auction).
Throughout the day, the prices of derivatives might be rarely correct.
Pricing relationships are going to be incorrect.
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 2 of 7)
Continuous auctions require complex
technological implementations at the exchanges
1
2. Frequent Batch Auctions
Complex implementations are (i) prone to technological errors, (ii) require significant implementation costs
Increased costs for exchanges & market makers (who would have to alter their systems significantly)
Since the market is frozen for the duration of an auction,
participants would not know fair valuation during the auction* *(with continuous auctions – the fair valuation would theoretically be invisible forever)
2
Participants would not know a definitive price that would definitely lead to a trade in the auction
3
Without information, retail traders & investors with urgency would send orders at desperate levels
to increase their likelihood of having a trade
I don’t know what would lead to a trade in the auction. Let me be safe & try to buy @ Rs 110 instead of Rs 105.
(increasing price of buy order increase probability of trade in auction)
Impact on investors & market -1
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 2 of 7)
2. Frequent Batch Auctions
Continuous auctions is complex
& does not satisfy urgent needs for liquidity.
Participation could move to simpler exchanges globally
4
Impact on investors & market -2
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 2 of 7)
30
2. Frequent Batch Auctions
2. Activity within the auction are hidden from market participants *
Given the above complexity, exchange auctions are one of the most manipulated events globally
* If activity during the auction is not hidden, then participants could try manipulating the auctions close to the end of the auction.
Unfortunately, if zero information about activity is provided, then market participants are totally clueless and that is dangerous too.
Because of the two challenges above, most exchanges provide a summary of information (say an indicative price) – but indicative prices could be manipulated
As demonstrated by manipulations in Euronext Brussels & Euronext Amsterdam - possibility of manipulation is even higher when there are two auctions simultaneously in two exchanges – possibly why existing auctions in NSE & BSE are unpopular
1. The market is frozen for the duration of the auction
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud -1
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 2 of 7)
2. Frequent Batch Auctions
In illiquid assets, manipulators could force some auctions to have low indicative auction prices using very small quotes.
Sellers looking at the indicative price might end up selling at the artificially displayed low indicative price
In the next auction of those illiquid assets, manipulators could then display a high indicative auction price using small quotes.
Buyers looking at the indicative price might end up buying at the artificially displayed high indicative price
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud -2
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 2 of 7)
2. Frequent Batch Auctions
For auctions to be non-manipulatable, the moment when auctions are closed should be random
If any market participant figures out the methodology of auction period closing time of an exchange,
then possibility of manipulation is huge
Since the benefits of figuring auction closing timestamps are huge, the likelihood of market participants trying to corrupt exchange
professionals to obtain this information is high
When systems are made complex, benefits of corruption increase – therefore possibility of corruption increases
As corruption can change the tide significantly in a complex system,
distrust in the system increases (even when there is no corruption !)
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud -3
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 2 of 7)
2. Frequent Batch Auctions
Auctions of same asset at two exchange simultaneously is grossly manipulate-able.
Please find examples of manipulations from Euronext Brussels & Euronext Amsterdam which were fined by regulators
Using one big buy & another big sell order, a firm can control the indicative auction price
Very close to end of auctions, the firm would cancel their big buy order on exchange 1
They would also place another buy order in exchange 1 at a slightly lower price & another sell order in exchange 2 at a slightly higher price
The firms’ buy order at the lower price gets traded in exchange 1.
The auction on exchange 1 would happen at a lower price
Likewise on exchange 2, very close to end of auction, the firm would cancel their big sell order to cause the auction to close higher. And get their higher priced sell order executed on exchange 2
They would set the same indicative price at both exchanges Impact on
investors & market
Impact on market makers
Possibility of misuse & fraud -4
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 2 of 7)
2. Frequent Batch Auctions
Impact on investors & market
Impact on market makers
About Proposal
Possibility of misuse & fraud
Overall assessment
Beneficiaries
Losers
• Foreign exchanges - if international investors shift to exchanges with simpler structures
• Indian Exchanges – technological costs of implementation • Market participants whose strategies react on micro events
& now need complete overhaul: market makers & HFT • Indian Funds & investors if volumes shift abroad
(Existing auctions in equity segment at start of day are anyway unpopular – making the entire market a series of continuous auctions might spread inactivity everywhere)
Summary
Overall assessment
• While this is an interesting proposal, this would introduce a lot of technological complexity and scopes for manipulation
• Pricing relationships could go awry and price discovery could be uncertain
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 2 of 7)
3. Random Speed Bumps or delays in order processing / matching
Impact on market makers
Possibility of misuse & fraud
Overall assessment
Impact on investors & market
About Proposal Proposal: Introduction of randomized order processing delay
of few milliseconds to orders.
Precedent: 1. Thomson Reuters, ParFX apply randomized pause to orders that can result in trades (20 – 80 milliseconds)
(Such OTC (over-the-counter) market places are designed to cater to banks – only banks can place limit orders & be market makers, banks can choose to neither honor their quotes nor honor their trades and can annul them. The
counterparty information is also shared after the trade. Moreover only orders that can result in trades are delayed to assist the market maker – i.e. banks)
2. TSX Alpha imposes randomized delay (1-3 milliseconds). (Through a differentiated pricing & structural model, this market place is
designed to attract retail flow. Speed bump is only on orders that hit existing quotes, and are designed to deter institutions – market makers continue to
trade against retail traders) 3. IEX exchange will introduce sub-millisecond delay to liquidity taking IOC
orders
Objective: Discourage latency sensitive strategies as such delays would affect HFT, but not deter non-algo order flow where delay in milliseconds is insignificant
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 3 of 7)
3. Random Speed Bumps or delays in order processing / matching
Similar to the proposed ‘minimum resting time’, the proposal will mean that: market makers will not be able to modify their quotes till the random speed
bump is over
When an event happens after a market maker posted his quotes
& before the random speed bump is over
The market maker will not be able to modify his quotes till the speed bump of
his existing quotes is over (orders can only be modified after an acknowledgement
from the exchange – proposed to be delayed by a random speed bump)
To protect themselves, market makers might start quoting wider.
Some might quit !
Liquidity taker
My sell quote is @Rs 120 instead of Rs 105
My buy quote is @ Rs 90
instead of Rs 104
Automated MM
Will sell @Rs 105 Will buy @ Rs 104
Automated MM
Event X (new fair price is 107)
Need to increase my quote prices to 106 & 108 Automated
MM
But I have to wait till end of random speed bump
This will be exploited by other market participants. The market maker will be hit
on the wrong side
Let’s trade against the market maker’s incorrect quotes
Automated MM
Sold @ Rs 105 for a loss
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 3 of 7)
3. Random Speed Bumps or delays in order processing / matching
However, market makers will respond by quoting wider & safer, therefore no one is going to benefit much
Traders who exploit forced delays on market makers will benefit (Often liquidity takers who send a lot of IOC orders hoping for misprices)
2
All effects of reduced market making as discussed (i.e. higher bid
ask spreads, higher volatility, higher impact cost, etc)
1
Institutional clients would have difficulties in benchmarking
execution strategies provided by brokers
– as performance variation would be attributed to randomness
3
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 3 of 7)
3. Random Speed Bumps or delays in order processing / matching
Beneficiaries
Losers
• Foreign exchanges - if international investors shift to exchanges with simpler structures
• Indian Exchanges – technological implementation costs • Participants whose strategies react on micro events &
now need complete overhaul: market makers & HFT • Market Makers • Institutional clients (difficulty in evaluating best
execution brokers)
Summary
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Overall assessment • Stated intention is to deter HFT – but most HFT are good (e.g. market
making; arbitrage that rectifies prices;etc). Police bad HFT, but don’t create hurdles for all HFT
• Short random bump might mean nothing for non co-located participants based in other cities
• Long random bump will harm market makers and will affect liquidity • Liquidity providing traders (i.e. market makers) might be replaced by liquidity
consuming traders - traders who do the complex task of calculating the fair valuation & providing liquidity will mostly quit. The suggestion of introducing speed bumps in IOC or liquidity taking orders could be explored
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 3 of 7)
4 Randomization of orders received during a period
Impact on investors & market
Possibility of misuse & fraud
Overall assessment
Impact on market makers
Proposal: Orders received during a predefined time period (say 1-2 seconds period) is randomized and the revised queue with a new time
priority is then forwarded to the order matching engine
Objective: nullify latency advantage of the co-located players to a large extent that they get on the basis of physical proximity to the trading
platform & thereby, discourage latency sensitive active strategies
Precedent: ICAP’s EBS Market Matching Platform has a random batching window of 1, 2 or 3 milliseconds
(The time scale of EBS’s randomization window is much lesser than the proposed time scale. And EBS is a FX matching platform with same philosophies as that of the examples discussed in the previous section – i.e. aiding traditional market
makers with weak technology)
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 4 of 7)
4 Randomization of orders received during a period
Current scenario: Market makers can modify their quotes in the market when events happen – and thus manage their risks
After the new proposal: By forcing randomness (that too for 1-2 seconds !), replace requests sent by
market makers to manage their risks could be processed much later than others
Despite responding to risks quickly, market makers would get hit on the wrong side
Will buy @105
To protect themselves, market makers might either start quoting wider.
Some might quit !
My sell quote is @Rs 120 instead of Rs 105
My buy quote is @ Rs 90
instead of Rs 104
Automated MM
Will sell @Rs 105 Will buy @ Rs 104
Automated MM
Event X (new fair price is 107.5)
Automated MM
Liquidity taker
Will sell @Rs 108 Will buy @ Rs 107
Automated MM
Sold @ Rs 105 for a loss
Market maker is assigned lower priority by random
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 4 of 7)
Lack of certainty in system could lead to lack of confidence in system
To increase the probability of a trade, traders would send multiple orders to do the same trade*
(*hoping one of the multiple orders would end up higher in priority)
4 Randomization of orders received during a period
All effects of reduced market making as discussed
(i.e. higher bid ask spreads, higher volatility, higher impact cost, etc)
1
Impact on market makers
Overall assessment
About Proposal
Because of randomization,
the probability of getting a trade reduces
2
The load on the exchange infrastructure is going to increase
High Frequency Traders who trade at top of order book will
never have certainty about their orders getting executed
3
2
Impact on investors & market -1
Possibility of misuse & fraud
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 4 of 7)
4 Randomization of orders received during a period
Because of increased uncertainty about execution of orders,
price disparity across financial instruments
will not be rectifiable by arbitrageurs
4
Since stock is trading at price S, the future should be priced at F=Sert
Arbitrageur
However, the prices of the stock & future are not in sync, I can try to trade both and
lock an arbitrage profit
Arbitrageur
Unfortunately, because of randomizations, I am not sure of executions in both my
orders – so I cannot attempt the arbitrage
Pricing inefficiencies will increase because of inability of arbitrageurs to rectify pricing mismatches
Impact on investors & market -2
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Arbitrageur
As speed advantage reduces,
firms will reduce technology investments (tech arms race between HFT firms will reduce)
5
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 4 of 7)
4 Randomization of orders received during a period
Reverse engineering the computer algorithm that randomizes the
orders is within the realm of possibilities
1
Once the randomization logic is known (or obtained through nefarious means), the possibilities for manipulation are limitless
Increased possibilities of manipulation through spoofing
2
Sending a lot of spoofing orders after someone else’s genuine order can cause a lot of orders to be ahead of the genuine order in priority
If there are more orders arriving in a particular time window, any genuine order will be randomly behind half of them
Randomization is applied to all orders entering the exchange within a particular time window
Currently, sending a lot of orders after a genuine order cannot modify the priority of the original genuine order
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 4 of 7)
4 Randomization of orders received during a period
Impact on investors & market
Impact on market makers
Overall assessment
About Proposal
Possibility of misuse & fraud
Beneficiaries
Losers
• Liquidity takers who send a lot of orders hoping to hit randomly delayed orders of market makers
• HFT firms: Reduced technological arms race
• Exchanges - technological costs of implementation • Exchanges - to handle higher order flow • Funds & investors – increased pricing inefficiencies • Market makers • HFT firms: Uncertainty of trades
Summary
• Nefarious HFT activity should be monitored and prevented, current proposals will hurt all HFT activity
• Randomization could help alleviate the arms race currently going on within HFT firms (at the same time curtailing innovation)
• On today’s date, if introduced, the randomization window should be of the scale of single digit microseconds and not more. However with time, sophistication in technology might render such thresholds irrelevant
Overall Assessment
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 4 of 7)
5. Maximum order message-to-trade ratio requirement
Proposal: Market participants to execute at least one trade for a set number of order messages sent to a trading venue.
Not be able to place such orders that further increase the ratio, after the limit is breached
Impact on investors & market
Possibility of misuse & fraud
Overall assessment
Impact on market makers
About Proposal
Objective: Increase the likelihood of a viewed quote being available to trade and reduce hyper-active order book participation
Precedent: Not aware of other precedents (of not being able to send orders after a certain ratio is reached)
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 5 of 7)
46
5. Maximum order message-to-trade ratio requirement
Currently algo market makers can provide buy & sell quotes which are very close to their fair valuation.
Because they can modify their quotes in response to events
46
After the new proposal: When events happen in the market
Market makers might not be able to modify their quotes (because of need to maintain the required message-to-trade ratio – which could be
impacted because others have not traded against them)
Because I have not traded enough, I cannot modify my quotes now
Automated MM
Liquidity taker
Let me trade against the market maker’s stale quotes which have changed after event X
The market maker will therefore be hit on the wrong side. To protect themselves, market makers might start quoting wider
As I cannot modify my quotes frequently, I will now sell @Rs 110 instead of Rs 105
I will now buy @Rs 100 instead of Rs 104
Automated MM
This might affect genuine investors who might have to pay a wider spread
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 5 of 7)
Technological infrastructure & data storage complexity
for market participants will reduce
Exchanges will require less complex infrastructure
3
The number of orders sent to the exchange will reduce
2
5. Maximum order message-to-trade ratio requirement
Market participants will have to process less market data
4
Market makers might stop providing liquidity in less liquid assets
5
Resolvable using different order order-to-trade ratio requirements for different types of instruments (classified based on liquidity levels)
All effects of reduced market making as discussed before (i.e. higher bid ask spreads, higher volatility, higher impact cost, etc)
1
Impact on investors & market -1
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 5 of 7)
5. Maximum order message-to-trade ratio requirement
Impact on investors & market -2
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Stopping the ability to modify existing orders (after the order-trade ratio limit is reached) will expose all active orders of that trader as
targets for exploitation when the market moves
6
While it appears too strict – but will force traders to be judicious while sending orders
Stopping the ability to modify existing orders (after the order-trade ratio limit is reached) will require real time monitoring by the
exchange – which will increase the processing time of the exchange
7
Increased exchange processing time will increase the time taken by the exchange to send acknowledgements to market makers
Delayed acknowledgements will prevent market makers from modifying their orders after events. Increasing the likelihood of the
market maker to be hit on the wrong side
This will lead to wider bid-ask spreads from the market makers
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 5 of 7)
5. Maximum order message-to-trade ratio requirement
Impact on market makers
Possibility of misuse & fraud
About Proposal
Impact on investors & market
Overall assessment
Beneficiaries
Losers
• Exchanges: Reduced order load • Exchanges & Market Participants: Reduced market data
infrastructure
• Liquidity takers who send a lot of mindless orders hoping to find misprices of others (typically market makers)
• Exchanges: To track order to trade in real time and stop new orders when the limit is reached
Summary
• A logical order-to-trade ratio (that is customized for different liquidity classes) is a positive step
• Real time monitoring to prevent orders after a certain limit might introduce excess complexity for the exchange - which could expose the system to failures arising from complexity.
Overall Assessment
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 5 of 7)
6. Separate queues for colo orders and non-colo orders
Proposal: Stock exchanges facilitating co-location / proximity hosting shall implement an order handling architecture comprising of two
separate queues for co-located and non-colocated orders such that orders are picked up from each queue alternatively
Impact on investors & market
Possibility of misuse & fraud
Overall assessment
About Proposal
Impact on market makers
Objective: Orders generated from a non-colocated space get a fair chance of execution and address concerns related to being crowded-
out by orders placed from colocation
Precedent: Not aware of other precedents
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 6 of 7)
51
6. Separate queues for colo orders and non-colo orders
51
Current scenario: When events happen in the market, market makers try to get this information quickly using colocation, and then adjust their quotes quickly to reduce their
risks. Other traders who have a need for quick information also use colocation.
Long term investors & hedgers who are not bothered about minuscule changes in the market will not use colocation servers
(e.g.: someone with a long term investment with a target of 10% will not be affected by 0.05% change in prices [price change can be either favorable or against]. But a market maker whose bid-ask spread itself is
0.05%, will be affected adversely if prices move (price change in either direction is unfavourable)
After the new proposal: Messages sent by market makers to the exchange to adjust quotes after events
will go behind in priority after orders from non-colocation servers who do not have genuine urgency
Will buy @110
Will sell @Rs 105 Will buy @ Rs 104
Automated MM
Event X (new fair price is 105.5)
Automated MM
Will sell @Rs 106 Will buy @ Rs 105
Automated MM Sold @ Rs 105 for a
loss
Message from non-colo is given preference
Long term investor
To protect themselves, market makers might either start quoting wider.
Some might quit !
My sell quote is @Rs 120 instead of Rs 105
My buy quote is @ Rs 90
instead of Rs 104
Automated MM
People who have a need for speed should use colocation. What needs to be analyzed is if the costs
are prohibitively high (and not possibly the proposal of giving non-colocation orders higher priority)
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 6 of 7)
52
6. Separate queues for colo orders and non-colo orders
52
Since orders from the colocation queue & the non-colocation queue are going to be processed alternately
Traders with a need for speed with try to be fast on both the queues
In addition to colocation, they would also put their servers at the non-colocation site nearest to the exchange
Increased infrastructure cost for other market participants who need to be quick & competitive
Many market participants might quit the market
Reduced liquidity
All effects of reduced market making as discussed before (i.e. higher bid ask spreads, higher volatility, higher impact cost, etc)
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 6 of 7)
6. Separate queues for colo orders and non-colo orders
Higher complexity in proposed solution
Higher likelihood of fraud
By sending a lot of spamming orders in one of the two queues (i.e. queue of colocation orders, or queue of non-colocation orders) …
… manipulators can significantly slow down genuine orders in one queue vis-à-vis the other queue.
Forced delays on one queue can lead to multiple ways of manipulation
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 6 of 7)
6. Separate queues for colo orders and non-colo orders
Higher complexity in proposed solution
Higher likelihood of fraud
By sending a lot of spamming orders in one of the two queues (i.e. queue of colocation orders, or queue of non-colocation orders) …
… manipulators can significantly slow down genuine orders in one queue vis-à-vis the other queue.
Forced delays on one queue can lead to multiple ways of manipulation (Please see subsequent page for an illustrated example)
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud -1
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 6 of 7)
6. Separate queues for colo orders and non-colo orders
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud -2
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 6 of 7)
Queue 1 (say non-colo) Queue 2 (say colo)
←Ti
me
Actual sequence of orders sent on both queues
Sequence in which orders will be processed (new proposal of 2 queues alternately)
Spam Order by Manipulator
Spam (manipulator)
Spam (manipulator)
Spam (manipulator)
Spam (manipulator)
Spam (manipulator)
Naïve trader (buy@10 sell @11)
Normal order
Normal order
Normal order
Normal order
Normal order
Manipulator looks at current fair price & hits novice trader
Spam Order by Manipulator
Normal order
Spam (manipulator)
Normal order
Spam (manipulator)
Spam (manipulator)
Normal order
Spam (manipulator)
Normal order
Spam (manipulator)
Normal order
Naïve trader (buy@10 sell@11)
Manipulator looks at current fair price &hits novice trader
At this point the manipulator knows orders
on queue 1 will be stale priced & will hit them (say
buy@10 if prices have increased to 12)
Spam orders
by manipulator to slow this
queue
←Ti
me
6. Separate queues for colo orders and non-colo orders
Impact on investors & market
Impact on market makers
About Proposal
Possibility of misuse & fraud
Overall assessment
Beneficiaries
Losers
• Data centre & real estate businesses that set up the fastest infrastructure to trade near the exchange
• Low latency traders - as they will have to set up parallel infrastructure in the fastest non-colo location
• Exchanges: technological cost of implementation
Summary
• The concerns raised by non-colocated market participants of being ‘crowded out’ do not make logical meaning. Irrespective of where they send orders from, they will be handled equivalently.
• People who have a need for speed should use colocation services. What needs to be analyzed is whether people with a need for speed are unable to use colocation because of prohibitive exchange costs
Overall Assessment
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 6 of 7)
7. Review of Tick-by-Tick data feed
Proposal: Exchanges should provide ‘Structured Data’ containing Top 20 / 30 / 50 bids / asks, market depth, etc. to all the market participants at a prescribed time interval (or as real-time feed).
Impact on investors & market
Possibility of misuse & fraud
Overall assessment
Impact on market makers
About Proposal
Objective 1: Tick by Tick data feed (i.e. real time feed) is not availed by small players because of cost concerns.
The proposal is to create a level playing field irrespective of their financial strength
Objective 2: Tick by Tick data feed (i.e. real time feed) is not availed by small players because it is data heavy. However, big
players who use this data format get additional data from which they derive additional insights. This causes disparity and
inequality.
The proposal is to create a level playing field irrespective of their technological strength
Precedent: Historically, because of technological complexities, most exchanges used to provide a snapshot of data at the end of a fixed
timeframe. Some exchanges in emerging economies still do. However most provide structured data upto very few levels
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 7 of 7)
7. Review of Tick-by-Tick data feed
Currently algo market makers get information of each and every event as they happen (tick by tick data).
They can then modify their quotes as quickly as events happen
After the new proposal: If the structured data is provided at a
prescribed time interval
Market makers will be in the dark about events that happen between
two data packets
Will buy @Rs 104.5
Will sell @Rs 105.5 Automated MM
Got sold at 105.5 Automated MM
Market data packet implying fair price = Rs 105
Market data packet implying fair price = Rs 107.5
Event X (new fair price is 107.5)
Market makers would get adverse trades during those dark periods
To protect themselves, market makers
might either start quoting wider. Or QUIT !
My sell quote is @Rs 120 instead of Rs 105
My buy quote is @ Rs 90
instead of Rs 104
Automated MM
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 7 of 7)
7. Review of Tick-by-Tick data feed
All effects of reduced market making as discussed before (i.e. higher bid ask spreads, higher volatility, higher impact cost, etc)
New technology infrastructure to be implemented by exchange & market participants
Reduced data to be handed by market participants (in case of active instruments).
However, for inactive instruments, on every tick the exchange has to transmit 50 or 70 levels of order book data
instead of just forwarding the tick
1
2
3 Impact on
investors & market
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal If structured data is provided after a fixed time interval
If structured data is provided as a real time feed
Market participants will not have to build their own order book from tick by tick data
1
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 7 of 7)
7. Review of Tick-by-Tick data feed
Since most market participants will be in the ‘dark‘ between two data packets
Manipulators can significantly effect prices during the ‘dark period’ before anyone could even realize or react.
Manipulators get a free hand to do anything during the ‘dark
period’ without arbitrageurs / other market participants responding
Impact on investors & market
Impact on market makers
Possibility of misuse & fraud
Overall assessment
About Proposal If structured data is provided after a fixed time interval
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 7 of 7)
7. Review of Tick-by-Tick data feed
Impact on investors & market
Possibility of misuse & fraud
About Proposal
Impact on market makers
Overall assessment
Beneficiaries
Losers
• Liquidity takers who would keep sending mindless number of orders between two snapshots
• Current participants not using TBT data & who can analyze the additional levels of data
• Market Makers
Summary
• What needs to be analyzed is whether prohibitive costs prevent desirous market participants from using tick-by-tick data ?
• Some traders might not have a need for quick data – but they might desire additional levels of order book data (only available in TBT data currently). There could be an additional data packet type with frequent snapshots of more detailed order book information. In parallel TBT data should persist for traders who have a need for quick updates
• Lowering the standards to cater to the strength of the weak players might not encourage innovation
Overall Assessment
Automated MM
Introduction (type of market participants, role of market makers) Analysis of SEBI proposals (proposal 7 of 7)
About the Author
The author started his career with one of the leading
automated market making firms globally – making his
small contributions in stabilizing markets in both Europe
an US during the Lehman crisis of 2008.
After the crisis, the author designed market making
systems for the Indian markets – helping increase
liquidity in erstwhile illiquid derivative contracts in India,
contracts that help hedgers reduce risks significantly.
Currently, the author is a partner in an automated market
making proprietary trading firm in India – making markets to
ease entry and exit opportunities in Indian assets – that will
hopefully make Indian assets more attractive to investors
The author is a frequent speaker at conferences across
US, Europe and Asia on topics centered around
algorithmic trading – having been invited by leading
global educational institutions, exchanges, etc to
seminars, conferences on related topics.