Daniel Fernandez at mechanicalforex.com has put up a good analysis of the qualities that profitable systems have in common when thinking about automated trading system development In this specific context, he explores what makes it more likely to have a system that performs well with a set of historical data (“in sample”) with subsequent data (either historical data that’s beyond the in sample data set, or using actual data going forward, aka “out of sample”). He finds that a sufficient number of trades, high linear regression coefficients, and high system quality numbers (see here and here for some discussion about SQN) all contribute to success when using one’s automated trading system in the market going forward. However, those characteristics are just a starting point, and he closes with this caveat:
It is also worth noting that the above does not imply that the problem is solved, on the contrary the above research opens up much more questions than it answers, also needing much more testing across other symbols to validate whether the results are true for different types of instruments. What I hoped to show is that there seem to be some significant correlations between in-sample and out-of-sample statistical results for trading strategies (at least the parameter-less price pattern systems created using Kantu) and that researching this can help us get an idea of what the “best method” actually is to device a trading strategy to tackle future market conditions. Finding what statistical characteristics should be maximised and what characteristics we should not worry about are one of the most important questions within this area of research.
As always, please read it all to get the details.