Here’s another great question from list member Glenn B. (used with permission). He’s asking about Amibroker, but this applies more generally to any backtest.
Glenn B.:
Should I include these in backtests? I know AmiBroker has built-in commission tracking, but I didn’t see much on slippage. Is it worth factoring in, or better to skip?
Dave:
You should include commissions in your backtests, but I don’t recommend including slippage.
Here’s why I don’t.
Commissions are 100% predictable – there’s no guessing – like death and taxes, they’re a fact of life.
For trading strategies with lots of trades, including commissions can change the optimization a surprising amount.
Slippage, however, is incredibly unpredictable.
Trying to model slippage is hard and not worth the effort, in my opinion. (Sometimes your trades will experience positive slippage – meaning you get a better fill than your backtest – try modeling that!)
Does that mean you should put your head in the sand and pretend slippage doesn’t exist? Of course not.
You should have a general sense of how much slippage would make your strategy unprofitable.
Knowing that before trading your strategy with real money will give you a quicker path to confidence once you go live. (For more info, listen to the discussion on “pre-mortems” in the podcast episode called Your Backtest Looks Great – What Could Possibly Go Wrong?)
Rather than trying to predict slippage on any particular trade, I prefer to treat the backtest as what’s theoretically achievable by the strategy.
Your live trading will, of course, fall short of that threshold by some amount, but that’s fine – your strategy shouldn’t require perfect execution to be profitable.
When measuring performance against the backtest, though, I prefer to compare against “perfect execution.”
Great question, Glenn B. – thanks for sharing with the group.
-Dave
