Automatic generation of trading rules adjusted to the market state

Abstract

The search for profitable trading strategies has been driving research efforts for decades. Among the many different approaches that can be used to design trading rules (such as Particle Swarm Optimization, Genetic Algorithms, Artificial Neural Networks and Fuzzy Methods), there are some based on evolutionary computation that are especially interesting due to three key features: the process of rule generation is automatic, the resulting rules are interpretable, and their structure is flexible. Since Allen and Karjalainen published their seminal piece on evolution of trading rules using Genetic Programming (GP), many authors have made related contributions either based on the same technique, or Grammatical Evolution (GE). Most of these contributions generate investment rules based on a combination of raw market data and technical indicators and, unlike related approaches that use genetic algorithms or evolution strategies to optimize predefined rules, these have the advantage of creating flexible structures automatically. A common limitation is that it is often the case that the approaches are static and do not take into account the structural changes of the state of the market. Given that this phenomenon is very prevalent in financial time series, the decision rules are commonly derived from market environments that do not hold in test periods. The problem of adjusting to structural changes is that we must choose between two opposite extremes: keeping the same model over time, or updating it constantly. Even though the second might seem, at least in principle, more appropriate, there is a possibility that the constant change in the model will have undesirable consequences due to transaction costs. The evolutionary process of GP/GE considers commissions throughout the period as part of the fitness function, and that makes it select rules that generate a limited number of signals. However, it is possible that a constant model update interferes with that endogenous control mechanism of the number of purchase and sale orders. This Tesis tackles with dynamic trading system solutions based on the use of ensembles and GE. The approach combines the possibility of changing the model as a reaction to changes in the price generation mechanism, with an inertia component that mitigates the consequences of overtrading. We also work with a different approach that is not based on ensembles but on a system that takes advantage of an internal hysteresis mechanisms that is part of the own models.