options pricing - fuzzy term paper

Upload: achalpremi

Post on 08-Apr-2018

226 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    1/21

    MAL 717

    Fuzzy Sets and Applications

    Paper overviewA Multiperiod Binomial Model for Pricing Options in a

    Vague Worldby

    Silvia Muzzioli and Costanza Torricelli

    Submitted by

    Achal Premi (2005MT50558)Sameer Jain (2005MT50445)Department of Mathematics

    IIT Delhi

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    2/21

    Contents

    S. No. Titles Page Nos.

    1. Introduction 31.1

    1.2

    Options

    Options Pricing

    3

    4

    2 Fuzzy Binomial Option Pricing Model

    (FBOPM)

    4

    2.1 Fuzzy Binomial Tree 4

    2.2 Risk Neutral Probability Intervals 6

    2.3 Pricing of an option in a vague world SinglePeriod Model 11

    2.4 Multi Period Model 13

    3 Defuzzification 15

    4 Proposed Extension 15 4.1

    4.2

    4.3

    Integration of fuzzy BOPM into fuzzy logic

    Justification for values of control parameter f

    Pattern Recognition & Similarity Measures

    15

    17

    17

    5 Conclusion 18

    Page 2 of21

    Prepared by: Achal Premi & Sameer Jain

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    3/21

    1. IntroductionThis paper gives an overview of the Binomial Option Pricing Model in a

    vague world published by Silvia Muzzioli, Costanza Torricelli in the

    Journal of Economic Dynamics & Control 28 (2004)and provides an

    extension to the model proposed by integrating the fuzzy pricing model

    with fuzzy logic rules and pattern recognition to capture the information

    about different states of an economy in calculating the price of an option.

    1.1. Optionsan option is a contract that gives someone the right to buy or sell an

    asset for a specified time at a specified price, but unlike a forward or a

    future contract, the buyer of the option is not under any obligation to

    exercise the option. The asset can be a real asset such as real estate,

    agricultural products or natural resources, or it can be a financial asset

    such as stock, bond, stock index, foreign currency, etc. There are two

    types of options, namely, call and put options

    Call Option:The buyer of a call option has the right but not theobligation to buy an agreed quantity of an asset (the underlying security)

    from the seller (writer) of the option at a certain time (the expirationdate) for a certain price (Strike price). The seller is obligated to sell the

    asset if the buyer wishes to exercise his right. Initially, the buyer of the

    option has to pay a certain amount to the seller for purchasing the option

    which is called the premium or the call price. Given below are the payoff

    and profit curves for the buyer and seller of a call option.

    Page 3 of21

    Prepared by: Achal Premi & Sameer Jain

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    4/21

    FROM BUYER'S PERSPECTIVE FROM SELLER'S

    PERSPECTIVE

    Put Option:The buyer of a put option has the right but not theobligation to sell an agreed quantity of an asset (the underlying security)to the seller (writer) of the option at a certain time (the expiration date)

    at a certain price (Strike price). The seller is obligated to buy the asset if

    the buyer wishes to exercise his right. Given below are the payoff and

    profit curves for the buyer and seller of a put option

    Page 4 of21

    Prepared by: Achal Premi & Sameer Jain

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    5/21

    FROM BUYER'S PERSPECTIVE FROM SELLER'S

    PERSPECTIVE

    1.2. Option Pricing

    Option pricing is an attempt to determine the fair market value of the

    premium that should be paid by the buyer to the seller of the option.

    (Here the term fair refers to a desirable condition that both buyer and

    seller should have equal expected returns from the option contract).

    The aim is to develop a model that takes into account as much market

    information as possible and come up with a fair value for the option

    price (call price or put price).

    The different models differ in their ability to cover different aspects of the

    market. Thus, a model can be considered to be good if it can take account

    of the future movements of underlying security with some precision and

    hence determine a fair premium value payable at the time of buying

    the option.

    2. Fuzzy Binomial Options pricing model

    (FBOPM)2.1. Fuzzy Binomial Treelet us consider a one period model where t {0, 1} is the time. Let P0 be

    the price of the stock at time t = 0 and P1 be the price at time t = 1, u &

    d be the up and down jump factors with the probabilities p and (1-p); p

    [0, 1].

    In the standard binomial option pricing model (BOPM), we assume u and d

    to be crisp values so that the stock value at time t = 1 is either

    P1= uP0 with probability p OR

    P1= dP0 with probability 1-p.

    But this assumption of the BOPM neglects the vagueness involved in the

    stock price movements in the real world and hence is not expected to

    yield a fair option price value. To counter this limitation the authors

    Page 5 of21

    Prepared by: Achal Premi & Sameer Jain

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    6/21

    have assumed u and d to be fuzzy numbers. For the sake of

    computational simplicity they are assumed to be triangular fuzzy

    numbers (TFNs).

    Fig 1: The two possible jump factors

    But How to get the jump factors u & d?

    Standard BOPM Analogously for the

    fuzzy BOPMWork done by Cox et al. (1979)

    leads to

    et

    u

    =

    ud e

    t1==

    , where

    - volatility of the underlying

    asset,

    t - length of the time period.

    Triangular fuzzy numbers u = (u1,

    u2, u3) & d=(d1, d2, d3) are givenby

    et

    u

    =1

    1

    , et

    u

    =2

    2

    ,

    et

    u

    =3

    3

    ( 321

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    7/21

    Where

    PT defuzzified option price calculated from

    the modelPM actual market price of the optionn number of observations from the past

    datar continuously compounded interest rate

    The second condition in the NLPP takes care of the no arbitragecondition (which says that no transaction or portfolio can make a profit

    without risk). This condition is also intuitively appealing because

    if etr > e

    t1then

    o no investor would wish to invest in the risky assets and

    would confine only to the risk free investments and

    if etr < e

    t 1

    o no investor would wish to make any risk free investment

    2.2. Risk neutral probability intervals

    We introduced p and 1-p as the probabilities of the up and down jump

    factors, but how do we get these probabilities. As for the pricing

    methodology, a risk neutral (fair) valuation approach is used therefore

    these probabilities should be such that they incorporate the fair (risk

    neutral) pricing strategy.

    The aim of this section is to derive these risk neutral probabilities in order

    to price a call written on a stock. Let us consider a one period model

    where t {0, 1} is the time

    Let pu & dp be the up and down jump probabilities. Then

    Page 7 of21

    Prepared by: Achal Premi & Sameer Jain

    up + dp

    = 1

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    8/21

    and from the idea ofrisk neutral valuation, we require

    1(000 rPpuPpdP ud +=+

    LHS = expected value of P1 (expected value of stock at t =1).RHS = the value of a risk free investment of amount P0 (invested at time

    t=0) at time t=1.

    Since u & d are fuzzy numbers, writing them in terms of cut we get

    (1)Solving these equations for a particular value of d

    [d

    1 ,d

    3 ] and a

    particular value of u

    [u

    1 ,u

    3 ] we get crispdp

    andup

    values. On

    varying d & u values on these intervals we get range of values forup

    anddp

    .

    We get maximum possible range ofup

    &dp

    values at a given

    cut

    by solving the following two systems of equations

    Solving system (2) yields

    Solving system (3) yields

    Page 8 of21

    Prepared by: Achal Premi & Sameer Jain

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    9/21

    The two solutions thus give the bounds on the probability intervals:

    It is easy to check that the following duality relations hold:

    up + dp

    =1d

    p + up

    =1Hence for a given -cut we get intervals for the possible values of up &

    dp . These intervals have a maximum spread at = 0 and reduce to crisp

    values at = 1. This is justified because it can be easily seen that

    derivative w.r.t is positive for both the left bounds and negative for

    both the right bounds of the probabilities.

    Also it is easy to verify the following claimSecond derivative

    positive condition

    Second derivative

    negative condition

    Second derivative

    zero condition

    up u3-u2 > d3-d2 u3-u2 < d3-d2 u3-u2 = d3-d2

    up u2-u1 > d2-d1 u2-u1 < d2-d1 u2-u1 = d2-d1

    dp u2-u1 < d2-d1 u2-u1 > d2-d1 u2-u1 = d2-d1

    Page 9 of21

    Prepared by: Achal Premi & Sameer Jain

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    10/21

    dp u3-u2 < d3-d2 u3-u2 > d3-d2 u3-u2 = d3-d2

    Taking use of the information in the above table we can have different

    shapes for curves of up and dp depending upon the relative positionsof u1, u2, u3 and d1, d2, d3 which are illustrated in tables below

    TABLE 2

    Page 10 of21

    Prepared by: Achal Premi & Sameer Jain

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    11/21

    Table3

    These tables do not depict the cases in which the second derivative of up ,

    up , dp , dp are zero when up & dp are TFNs.

    By inspection of the tables 2 & 3, it is clear that the shape of the risk

    neutral probabilities depends on the relative positioning of u & d. In

    particular if we fix the two peaks, d2 and u2 of the two jump factors, from

    table 2 we can see that when the distribution of u is closer to that of d i.e.

    u & d are less distinct, then both up & dp are closer to a crisp number.

    Conversely, when u & d are more distinct, up & dp are more vague

    Page 11 of21

    Prepared by: Achal Premi & Sameer Jain

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    12/21

    2.3. Pricing of an option in vague world

    single period model

    Having obtained the risk neutral probabilities up & dp in the previous

    section we now use them to price an option on a stock. We only considerthe one period model in this section.

    At the time of expiry a call option has a positive value (payoff) if the priceof the underlying stock at that time is greater than the strike price and iszero otherwise. Now in our fuzzy model at time t=1 the stock price iseither P0d or P0u, which are triangular fuzzy numbers (because u & dare triangular fuzzy numbers.)

    To consider an interesting contract let us assume that the strike price (X)

    is between the highest value of stock in the down state and the lowest

    value in the up state i.e.

    Let us denote

    C (u) =the payoff of call in the up state

    C (d) = the payoff in the down state.

    Then clearly

    C (u) = (P0u-X) = (P0u1-X, P0u2-X, P0u3-X)

    C (d) = 0

    Since u is a TFNC (u) is also a TFN which is given in the -cut

    representation as

    Now by the risk neutral valuation strategy we can determine the call price

    C0 at t=0 by

    Where

    E refers to the expectation under the risk neutral probabilitiesC1is the call payoff at time t=1

    This equation signifies that

    The discounted value of expected call payoff at time t=1 equals

    the value of call at time t=0.

    Page 12 of21

    Prepared by: Achal Premi & Sameer Jain

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    13/21

    condition that the expected returns of the writer and the buyer

    should be same at the time of expiry which is precisely the ideology

    behind risk neutral valuation.

    Since Cd = 0 in our case therefore

    The rules of multiplication between fuzzy numbers then lead to

    It is easy to prove that as increases this interval size of call option

    price shrinks.

    In particular for =1 the interval reduces to a single crisp value

    which is the same result as in standard BOPM (crisp case) with u2 and d2

    as up and down jump factors as expected.

    Also it is easy to verify the following claims

    Second derivative

    positive condition

    Second

    derivative

    negative

    condition

    Second

    derivative zero

    condition

    0

    C (u3-d3)(p0u2-X) >

    (u2-d2)(p0u1-X)

    (u3-d3)(p0u2-X)

    (u1-d1)(p0u2-X)

    (u2-d2)(p0u3-X) 1

    Neutral f 1

    Depression f < 1

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    19/21

    These f values may be decided from the expert knowledge of the

    investors, after carrying out extensive research on the historical data.

    4.2. Justification for values of Control

    Parameter f

    As an example consider the expression for C0 interval in the one period

    fuzzy BOPM

    This can also be considered as

    C0 =

    +

    + )

    11(),

    11(

    rr

    where , , , are some constants for a given value of and R =

    1+r.

    Clearly to increase C0 in the booming economy by using R* instead of

    R we need to have R* > R which in turn implies f >1. Similar observations

    can be made for the other economy states.

    But the question which arises now is how you would identify the state of

    economy at any particular point of time. This can be achieved through the

    concept of fuzzy pattern recognition which in turn uses similarity

    measures.

    4.3. Pattern Recognition & Similarity

    Measures

    Page 19 of21

    Prepared by: Achal Premi & Sameer Jain

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    20/21

    Let at any point in time the jump factors as interpreted by an investor

    be u & d (TFNs). Now we wish to know the state of the economy to

    which these jump factors resemble the most. For this we need the

    concept of similarity measures.

    Let B = (u,d) be the new fuzzy input (consisting of two fuzzy numbers:

    u and d)

    E1, E2, E3 be the three fuzzy classes each having two features Ei1 & Ei2,

    where

    Ei1 refers to up jump factor in Ei.

    Ei2 refers to down jump factor in Ei.

    Define the similarity measure between B & Ei as

    [B,Ei] = (u,Ei1) + (d,Ei2) , where

    (u,Ei1) =2

    1(u 1iE + 1iEu )

    u 1iE = xmax min { )(),(

    1xx

    Eiu } = inner fuzzy product

    between u & 1iE

    u 1iE = xmin max { )(),(

    1xx

    Eiu } = outer fuzzy product

    between u & 1iE

    Let max [B, Ei] = [B, Er] 1i3then we say that the current market state resembles Er state the most

    Now once we know the current market state by the above pattern

    recognition procedure, we take the corresponding control parameter value f

    and use the modified risk free rate of interest R* in the fuzzy BOPM to

    calculate the option price which is expected to capture a broader knowledge

    of the market as compared to the earlier fuzzy BOPM.

    5. Conclusion

    This paper tells about the basic aim and strategy involved in developing

    any Option Pricing model and provides an overview of the Risk Neutral

    valuation strategy for pricing options in a vague market published by

    Silvia Muzzioli and Costanza Torricelli in 2004.

    Page 20 of21

    Prepared by: Achal Premi & Sameer Jain

  • 8/7/2019 Options Pricing - Fuzzy Term Paper

    21/21

    But this existing model doesnt take into account the role of overallmarket sentiments in an investors decisions at any point of time. Thisincapability is partly due to the use of risk free rate of interest directlyinto the model, whatever may be the state of the market. Thisassumption is a valid one only when the expected returns from riskyinvestments are comparable to those from risk free investments whichis not the case in skewed economy states like when the economy isbooming or in depression.

    Thus this paper review tries to overcome this incapability by integratingthe existing model with fuzzy logic and fuzzy pattern recognition toidentify the state of economy at any point in time and consequently usea modified risk free rate of interest (R* = Rxf) while calculating the

    premium value. This approach is intuitively more appealing and is thusexpected to give better accuracy than the existing model.

    A further possible extension could be to use these steps on past dataand validating these results

    P 21 f 21