Optimization Of Trading Rules In The Crude Oil Futures Market With Agent-based Simulation

The crude oil futures market plays a critical role in energy finance. Choosing suitable strategies to increase investment profit is an important issue for investors in the crude oil futures market. This paper proposes an agent-based simulation approach to generate optimal trading rules in crude oil futures markets. With this approach, the traders can adjust their trading rules dynamically based on their performance during investments cycles. The main work and conclusions of this paper are as follows:(1) Based on the complex adaptive system theory, an agent-based model was constructed to simulate the crude oil futures market trading process. Existing studies about trading rule optimization always use static history price and neglect the influences of the investment behavior. Agent-based approach can model this feedback mechanism. Three are three main adaptive “Agents” in our model: traders, settlement center and the market administration. We design their decision rules and the interaction mechanisms between different market agents according the actual market transaction. Traders make investment according a market information pool containing different types of basic trading rules.(2) To adaptive market changes, genetic algorithms are used to evolve the trading rules set of each traders in our model, thus the traders can achieve dynamic accumulation of experience in the investment process. A trading rules set and dynamic evaluation mechanism give traders the ability to select optimal rules according to their performance during an investment cycle. Based on their credit distribution, traders in our model can create new trading rule and eliminate inferior rules, improving their strategies constantly.(3) A crude oil futures market simulation system was developed based on the SWARM platform to conducted simulation experiments. We use moving average trading rules, MACD rules, WMSR indicators and fundamental rules to design the basic trading rules set. The history crude oil futures prices and the spot prices in 30 years are used to conduct 5 group experiments in different market circumstances. The simulation model help find the best combinations of the basic rules and give investment advices in different market circumstances. Price trends and the statistics feature of the return rate of the simulation results are consistent with the real crude oil futures market.(4) The results show that our approach could help traders make profits in the crude oil futures market and find suitable trading rules. The average return rates of investors increase with time, proving the adaptive mechanism based on genetic algorithms can help improve trading rules. Using the decision-making mechanism we designed in our model, the basic technical trading rule performs better on forecasting increases in the market. WMSR indicators have best performance in total and moving average rules are better than MACD in forecasting increases. Results of comparative experiment show that traders will face high risk if they adjust trading rules too frequently. Increasing of the number of investors will lead more obvious fluctuations in market prices and the average return rates. These results can give some suggestion for traders when making investment in different circumstances. Totally, the approach we designed using agent-based simulation is proved helpful for traders can give some useful information on choosing trading rules in the crude oil futures market.