Backtesting and commissions

[LeeDanello]

I was wondering about the commissions used in backtesting. Reason being is that over the last 20 or so years commissions are becoming cheaper so backtesting a system using today’s low commissions will distort the system metrics. When we see our systems having only one or two losing years over the last 20 years because of these low commissions, for the shorter term systems they might in fact be losing years due to the higher transaction costs back then. Maybe we should be employing shorter backtest periods to reflect the reality, otherwise the metrics that are produced will be distorted. Case in question are the min $1 per trade via IB. Just wondering what others think?

[Nick Radge]

Well that may be true that a current system would be invalidated using old time comm rates, but that’s no irrelevant. The system is the system and the current comm’s are the current comm’s. If the system makes profits back in the 1995 – 2000 time on current comm’s then the system itself is robust.

If you’re concerned then go back to those earlier days and just use a higher comm rate.

Then the question is, “so what”?

Have you started trading your system Maurice?

[LeeDanello]

Nick I’m still paper trading it. Probably do so for another month. I will post up some results in the diary in the next couple of days.
I thought I’d ask the question because I read on the “other forum” about who had data back going back more than 20 years and what relevance that had on testing. So just thinking about that and testing strategies that far back. It was more for curiosities sake and how today’s commissions would those distort results.

[Author]

Posts