Adrian Reid explains how to choose commission and slippage assumptions for backtesting trading systems. He emphasizes matching backtest slippage to realistic, market-specific expectations (e.g., tiny for liquid ETFs like SPY, much larger for low-priced small-cap stocks), and using liquidity filters or recorded real-life slippage to inform settings.
The question is, when you're back testing, how much commission slippage should you allow for to make sure it's realistic and you're getting results that you can reproduce in real life?
And the answer actually depends largely on the system and the markets that you're trading. So what you want to do, the primary principle is to have a level of slippage and commissions in your backtest that represent what you would get in real life. So if you had a system that traded Australian small cap miners and there were stocks that were all under $0.05 or $0.10 and they were turning over a couple hundred thousand dollars a day on average, the amount of commissions and slippage that you have to, that you should include in your backtest is way more than if you were trading a mean reversion system on the SPY etf the us Right. Just worlds apart. So think about it in those terms is how much would I reasonably expect to get if I was following these rules on this universe? And it could be from as little as basically no more than the commission you pay to the broker, you know, 0.08% or something.1% or it could be a couple of percent.
In the case of a very small cap system.
And somewhere in between. Also with your liquidity filter, the higher the liquidity requirement for the system, the lower the slippage you're likely to get. So, you know, if you're, if you're trading trend following and your daily turnover requirement is $250,000, you might want 0.75 or 1% slippage and commission. But if your liquidity requirement is a million and a half, $2 million a day, then it could be as little as a quarter of a percent or 0.2.
So that's, that's the way I think about it. And if you don't know for sure, track the slippage you're actually getting on your trades and use that to inform you.
My final principle on this is I always like to overestimate it in the backtest. Yeah, okay. Because I don't want to design a system and then have that system collapse or be much worse than I expect because the commission is higher than I expected.
