«Back to research topics list

Applications in Finance

Papers on applications of evolutionary optimization in finance concern two main topics: portfolio optimization and trading rules optimization.

Selecting Best Investment Opportunities from Stock Portfolios Optimized by a Multiobjective Evolutionary Algorithm
The research presented in a paper at the GECCO 2015 Conference.

Highlights of the paper:
  • A multiobjective evolutionary algorithm was used for obtaining Pareto-optimal portfolios.
  • An analysis was performed in which historical and future performance of portfolios was compared.
  • Several investment strategies were tested which used optimized portfolios.
  • It was determined that a strategy that decides between high-return (but also high-risk) and low-risk (but also low-return) portfolios using historical behaviour of the stock market index is more effective than other simpler investment strategies.

Comparison of historical and future performance of portfolios.

Results obtained by investment strategies.



Analysis of Dynamic Properties of Stock Market Trading Experts Optimized with an Evolutionary Algorithm
The research presented in a paper at the EvoStar 2014 Conference.

Highlights of the paper:
  • Trading experts were built from individual trading rules, which produce buy/sell decisions.
  • An dynamic evolutionary algorithm was used for optimizing weights of the rules and decision thresholds.
  • Reinitialization of the popualtion and two strategies of introducing random immigrants were used for preventing excessive convergence of the population.
  • The experiments were performed on 50 stocks and ETF shares on a period of 76 weeks.
  • The best results were obtained when random immigrants were introduced to every generation of the evolutionary algorithm.
  • Usage of the rules was presented with respect to the time interval.


Results obtained for individual assets.

Usage patterns for trading rules obtained when random immigrants were introduced to every generation of the evolutionary algorithm.



Usage Patterns of Trading Rules in Stock Market Trading Strategies Optimized with Evolutionary Methods
The research presented in a paper at the EvoStar 2013 Conference.

Highlights of the paper:
  • Trading experts are built from individual trading rules, which produce buy/sell decisions.
  • An evolutionary algorithm is used for optimizing weights of the rules and decision thresholds.
  • Frequent itemset analysis is used for finding trading rules that coexist in well-performing trading agents.
  • Frequent itemsets present in no less than 50% of the trading experts and containing up to five trading rules were found.



Evolutionary Approach to Multiobjective Optimization of Portfolios That Reflect the Behaviour of Investment Funds
The research presented in a paper at the AIMSA 2012 Conference.

Highlights of the paper:
  • The paper addresses a problem of constructing portfolios from stocks and currencies with the goal of replicating the behaviour of an investment fund.
  • A mutation operator dedicated for portfolio optimization has been proposed, that modifies portfolio weights in such a way that the portfolio becomes more similar to the target investment fund.
  • A local search procedure which tries to generate solutions better with respect to two objectives based on solutions improving each of the objectives separately.
«Back to research topics list