Cheap Bias Estimates
On page 121 we’ll present a detailed examination of training bias, and on page 286 we’ll see a powerful way to deal with this serious problem. But for the moment, we’ll provide a rough overview of training bias and show how, if one has trained a trading system using differential evolution or some similar stochastic algorithm, we can get a rough but useful estimate of training bias as an inexpensive by-product of parameter optimization.
As we embark on developing a trading system, we have in our possession a set of historical data on which we will ...
