Returns on stocks have traditionally been modelled by fitting a suitable statistical process to empirical returns. Studies on agent based models of stock market have been carried out by researchers, primarily on US markets. This paper analyzes the empirical features generated using historical data from the Bombay Stock Exchange (BSE), employing the concept of agent based model proposed by LeBaron[2,3,8]. Agent-based approach to stock market considers stock prices as arising from the interaction of a number of individual investors. These investors are modeled as intelligent agents, using differing lengths of past information, each trading with its own rules adapting and evolving over time, and this in turn determines the market prices. It is seen that the model generates some features that are similar to those from actual data of the BSE.
General TermsAgent Based Modeling, Financial Forecasting, Neural Networks, Genetic Algorithm.
Key WordsAgents, Assets, Bombay Stock Exchange, Returns, Risky Assets, Risk Free Assets, Feed Forward Neural Networks, Rational Expectations Price, Forward Testing.
AGENT BASED MODELLING (ABM) OF STOCK MARKETS
IntroductionReturns achieved on stock markets contain certain characteristic features [1]. These features include a distribution of returns that is more peaked than the Gaussian distribution, periods of persistent high volatility, and correlation between volatility and trading volume. It has been shown that agent based models are able to demonstrate this, unlike the traditional financial models. The traditional economic models generally use either a simple distribution of returns such as the Gaussian and treat extreme events as outliers, or construct a statistical process which reproduces some of these features. The agent-based approach considers a population of intelligent adaptive agents and let them interact in order to maximize their financial performance. It has been shown that such an approach can replicate features of real stock markets [2,3,4,5,6,7,12,13]. This paper is an extension of the work of LeBaron and aims to study the Bombay Stock Exchange by forward testing [8] an Agent Based Model where a market is run using real data as the price input up to the current date, and then allowed to continue on into its future to enable the study of the empirical features. An attempt is made to study the behavior of agents with varying memory to see whether all horizon agents dominate the BSE market, preventing the stable long horizon agents to play a crucial role, as was established for the US market by LeBaron[2,3,8,9]. In this study, the financial data of BSE from the years 2003 to 2009 is considered for building the model, as against LeBaron"s model, where the data is entirely generated.
Why use an agent-based model for the stock market?Agent-based simulation is a bottom-up system approach to forecast and understand the behaviour of non-linear systems [5]. Interaction between agents is a key feature of the agent-based systems. As an alternative to regarding stock prices...
