Accounting for trading costs is in my book the top failure of most technical analysts: If you have a signal that generates a high zero-cost return, but trades so often that you definitely incur a lot of commissions, do the costs eat up all of the returns?

Because of the highly nonlinear structure of brokerage/commission costing systems there is often no “analytic” way to arrive at an estimate of after-cost returns given just the costing system and the raw before-cost returns, very often because the way in which the commissions eat into the trading capital will limit more aggressive positions, and also simply because before-cost returns do not penalize high-frequency signals relative to lower-frequency ones.

So, we model them.

Below we see the impact of trading the STI using the MACD long-only with and without costs, from a starting capital of $7000 in ...