Hybrid Methods of Some Evolutionary Computations AndKalman Filter on Option Pricing

International
OPEN ACCESS Journal
Of Modern Engineering Research (IJMER)
| IJMER | ISSN: 2249–6645 www.ijmer.com | Vol. 7 | Iss. 7 | July. 2017 | 40 |
Hybrid Methods of Some Evolutionary Computations
AndKalman Filter on Option Pricing
1
Joseph Ackora-Prah, 2
Emmanuel Owusu-Ansah, 3
Pearl AsieduwaaOsei
Kwame Nkrumah University of Science and Technology.
Corresponding author: ackoraprah@yahoo.co.uk
I. INTRODUCTION
Optimization is a mathematical technique through which best possible results are obtained under the
given set of conditions. Some of the methods used in optimization to address option problems are the traditional
and Evolutionary Computation methods. Option is one of the important research areas in the financial world.
Investors are provided with a world of opportunity through options, since they are given the power to adjust
market situations. These investors can choose to protect a particular asset by hedging to minimize risk. A lot of
people including investors, traders, and financial institutions are still searching for an optimal way to make
maximum profit. Many financial institutions recently use options because of its higher returns. Xu and
Zang(1975), used Kalman Filter to compute the stock prices and concluded that the method was fast and easy.
Kumar and Mansukhani(2011), also used the same algorithm in a situation where the estimate was linear and
discrete. The method removed the uncertainty while still maintaining vital information. Kalman Filter gave very
good predictions for the prices of stocks.Zheng(2015), used genetic programming in a situation where some
genetic operators were changing within the algorithm. Crossover and mutation were varied during the runs in an
attempt to solve a real world option pricing problem. The varying genetic programming method improved
performance of calculated option price. Chidambaran et al.(1998), developed a procedure upon which genetic
programming was applied to compute for option price. A given set of data was used to test for efficiency which
proved be good. Their model showed that the approximated true solution is better than the Black-Scholes model
when stock prices followed a jump-diffusion process. It was proved that the program out-performed other
models in many different settings due to it robustness and efficiency.
Investors are therefore confronted with the challenge of finding the optimal option price which may be
trapped in a local optimum (Ackora-Prah et al., 2014). In this paper we investigate three hybrid methods on
option price for a European put.
II. THEORETICAL CONCEPT
European Put Option. This provides the person the right to buy an underlying asset at a maturity time at a
strike price K, but not obliged to do so. ItsPayoff PTis:
ABSTRACT: The search for a better option price continues within the financial institution. In
pricing a put option, holders of the underlying stock always want to make the best decision by
maximizing profit. We present an optimal hybrid model among the following combinations: Kalman
Filter-Genetic Programming(KF-GP), Kalman Filter-Evolutionary Strategy(KF-ES) and Evolutionary
Strategy -Genetic Programming(ES- GP). Our results indicate that the hybrid method involving
Kalman Filter-Evolutionary Strategy(KF-ES) is the best model for any investor. Sensitivity analysis
was conducted on the model parameters to ascertain the rigidity of the model.
Keywords:Evolutionary Computation, Kalman Filter, Option Pricing.

Hybrid Methods Of Some Evolutionary Computations And Kalman Filter On Option Pricing
The seller of European put option will always expect the underlying asset priceto fall below the strike
price at the expiry date. The investor exercises atmaturity with an intrinsic value Pt, which will help in
calculating the optionprice as:

Kalman Filter.The invention of Kalman Filter has seen an extensive re-search area and has become
appealing due to its simple and robust nature.This method is a recursive mathematical process that fuses model
output withobserved data to estimate the hidden variable. It is an optimal estimator sinceit minimizes the mean
square error of the estimated parameter, which is knownto be a variance minimizing algorithm that always
update the state estimatewhen measurement becomes available by minimizing the trace of the error co-variance.
If the state space is linear then the Kalman scheme can be used forstate estimation thus:

Evolutionary Strategy. This method involves the random change of experimental setup. This experimental
strategy led to good result when it was tested.The major quality of Evolutionary Strategy is its ability to
incorporate majorparameters of the strategy such as standard deviation and the correlation coefficient
(covariance) of a normally distributed mutation.
Genetic programming. Genetic programming (GP) begins with a lot of randomly created computer programs.
This population of programs is progressively evolved over a series of generations. The evolutionary search uses
theDarwinian principle of natural selection (survival of the fittest) and analogs ofvarious naturally occurring
operations, including crossover (sexual recombination), mutation, gene duplication, gene deletion. Genetic
programming worksin an automated environment for creating a working computer program from
a high-level problem statement.
Graphical Simulation of the Underlying asset. The values for the simulation are as follows: underlying asset
price S0 = 100, interest rate = 0:15,volatility = 0:35 and the maturity time T = 3. The graph below shows
thesimulation of the underlying asset price over the given time [0; 3]: From thegraph it can be seen that the price
of the asset changes with time. The priceof the asset assume positive values as shown in the graph.

III. Methodology
vi) Mutation to generate the fittest offspring.
vii) Stopping criteria.

3.3. Genetic Programming and Evolutionary Strategy Algorithm. Thestock price are from the log form of
stock price equation using Geometric Brownian Motion model. The first part of this Hybrid solution was from
GeneticProgramming for selection and crossover until the new population was generated. Evolutionary Strategy
is then applied during mutation. The new mutatedpopulation is used to compute the option price of a European
put option. Thesolution to the hybrid method is obtained after the stopping criteria has beenmet. Simulations
was done using python 2.7 below;
4. Conclusion
Hybrid, Kalman Filter and GP gave GH ¢36:57, Kalman Filter and Evolutionary Strategy gave GH ¢27:03 and
Genetic Programming and EvolutionaryStrategy gave ¢34:85 as the value of the option price. Hybrids gave us
threesolutions from which we obtained our global minimum option price, which wasKalman Filter and
Evolutionary Strategy. It is therefore advisable for investorsto resort to our method in the option market. This
solution will help investorsto position themselves in the option market by reducing the loses that sometimes
occur.
REFERENCES
[1]. Yan Xu and Guosheng Zhang. Application of Kalman Filter in the prediction of stock price,Beijing Institute of
graphic communication China. 1975.
[2]. Prem Kumar, L. S., and SubirMansukhani,Prediction using KalmanFilter,Innovation and Devlopment, Mu Sigma
Business solution. 2011.
[3]. Zheng Yin, Anthony Brabazon, Conall O' Sullivan and Michael O'Neeill.Genetic Programming for dynamic
environment, Proceedings of the interna-tional multi conference on computer science information. 2015, 437-446.
[4]. Chidambaran N. K., Lee C-W. J. and Trigueros J. R., An adaptive evolu-tionary approach to option pricing via
genetic programming, Conference oncomputational Intelligence for Financial Engineering. 1998.
[5]. Joseph Ackora-Prah, Samuel Kwame Amponsah, Perpetual SaahAndamand Samuel Asante Gyamerah. A genetic
algorithm for option pricing:The American Put Option, Applied Mathematical Sciences.8(65):3197-3214
HIKARI Ltd
[6]. Sandra Mau. What is Kalman Filter and how can it be used for data fusion,Robotics Math. 2014, 16-811.
[7]. Mcwilliams, N. Pricing American options using Monte Carlo simulation,Master's Thesis, Summer Project. 2005.
[8]. Lamberton D., Lapeyre B. Introduction to stochastic calculus applied to_nance, Chapman and Hall. 1996.
[9]. Heigl A. Option pricing by means of genetic programming, Master's thesisTechnische University Wien. 2007.

Hybrid Methods of Some Evolutionary Computations AndKalman Filter on Option Pricing

More Related Content

More from IJMERJOURNAL (20)

Recently uploaded (20)

Hybrid Methods of Some Evolutionary Computations AndKalman Filter on Option Pricing