I think this gives backup to the phrase, "it's not so much the system you use as much as the strict following of a system" with position sizing, risk management, etc. I have been trading a system for about a year. Backtesting was done and looked very well. Very happy with the test results and also very happy with the actual results. I have been creating a new system, using this old system as a template and realized there was a bug in my original system that I would have not traded. However, I did do extensive back testing, so that was fine. Anyway, so interesting that it was not so much that I programmed with an error, as the system still had good trading rules and still made me $$$$. (LOL). I don't attribute this to luck, as I did extensive back testing and the mechanical system was followed religiously. Maybe you disagree... LOL. Larry
I once traded a system, not known to me at the time it was pretty breakeven over the long run, had up years and down years. Made very little money overall. But i did not have enough back test data at the time to know it. I started traded it with discipline and by luck it was one of it's up years and it made a lot of money. Then it started to drawdown and i stopped using it. Not because i knew it was a bad system but because i lost discipline to continue to follow it. Ignorance is bliss sometimes in trading. Especially when starting out as a trader.
Recently a couple of trades, on different days, I reversed by doubling trade the other direction, only to forget for a 10-15 seconds that oh chit 'I'm back in" the other way!!! Both worked out. Small gains. No such thing as luck. I definitely don't count on it and can do without it. As for system, good, bad or otherwise needs lots of trades to be valid.
I'm sure another trader following the same system would have lost his shirt. So you're lucky after all.
Schizo, You could be right. I hope not. This is a mechanical system, so not discretionary. The backtesting was good. My other system I have traded for 19 years and is also mechanical, but I do try to enter on a good entry point. Larry
Since a market is a random walk that can't be predicted, automated algos gain an edge at brute force. A discretionary trader would have not made a profit with that system but a machine can do it. You had a bug in your system from your point of view, as you tried to model it. But for that market it is not a bug, it is simply another group of trades that can succeed. One of my favourite approaches with algos, if the net result is negative, is to apply a random modifier at an early stage to suppress trades, let's say "don't trade 30% of the time". That can turn it into a profitable one. For that reason I never try to win with an algo, it is way better to try to minimize loses so the wins will come out of luck.
What I try to do is minimize my system rules and not optimize. Systems are based on business logic. With past system I did not optimize at all and got good results trading with the base system backtested going back many years. For me, this seems to help the system be robust when I go live.