Lost all faith in back tester

I will prelude this by stating that I did not put much stock in the MT4 back tester to begin with. Hopefully this will serve all who read it.


I was just sitting here preparing for the market to wake up later today mainly to continue testing my EA's and I was just trying different things and thought, let me write something that trades on a cci cross for the fun of it.. so i did.

One of the things that I have been getting in the habit of doing is whatever I code something, I also code the very opposite and test it along side because in the real world one just never knows. So, having said that... I coded my simple EA to go long when the cci(7) crosses the cci(14) on its way up. And vise versa for the short. OK so far so good. Against my common sense, I also wrote the opposite in a shadow of the original EA for the sake of being thorough. I back tested the first ea. {buy on cci(7) crossing cci(14) on way up.} It did great. all the nice numbers we'd all would love to be real. Then I tested the other one...

I was disappointed to see that it also did great! The second EA has 100% opposite logic as the first one. Yet they both do great on back test?!?!

hmmmm........

Any shred of faith that I may have had in the MT4 back tester has just flown out the window; lost in the cold winter wind.

Hal

I think you are missing the bigger picture here, trading system results are a cumulation of signals & money management, reversing signals will not produce an inverse backtest result. Adding to this, as we all no doubt are aware, there are many pitfalls in the backtester. If you have not fallen into one of these many pitfalls the backtester is acutally telling you something very important...

Hi,

I would go through the trades generated by both backtests and manually compare them against the data and see if they are accurate trades generated by the backtesting.

It could be for the parameters, timeframe and period of data tested, the results were accurate ... or not. But first check before dismissing them. If they were not accurate then you may be able to find ways to improve the backtesting accuracy.

Best regards
Alan

Thanks Craig/firehorse,

I have these 2, along with a bunch of others, lined up for forward testing on demo accounts later today so we will see what reality has to say.

You peak my curiosity; please elaborate...

The trading signal (and it's inverse) is statistically insignificant and/or the money managment code has a far larger impact on the result than the signal.