Question on forward testing

Hi all,

After 6 months studying forex I have finally developed my own system that really suits my style of trading.
I have been backtesting it manually (since I don't know how to create the proper EA) using one year data and the result is quite good by my standards.
Success rate: 65%
Profit to Loss ratio: 2:1 to 4:1

I am now in the phase of forward testing it using a demo account.
My question is that, how long should I test this system before I can safely say that it is profitable in the long run and it is okay to go live?

FYI, my system usually generates between 2 - 3 entry signals every day.

thanks for your help.

If you conducted your testing in an objective manner and acheived realistic results over a sample of 400+ trades there is no reason to believe that your strategy won't continue to perform well in the near future. In this case, my suggestion would be to start small and trade live. Record your results and compare them to your backtest to gain confidence that the backtesting was legitimate.

If you optimized your system in any way during the backtest then you definitley need to perform a walk forward analysis to determine the robustness of your sytem on unseen, or out of sample, data. While there is no single answer, many optimze on 70% of their data and walk forward on the remaining 30%. The results from the unseen data should be at least 50% as profitable as the known, or in-sample, data when compared on an annualized basis.

My question is that, how long should I test this system before I can safely say that it is profitable in the long run and it is okay to go live?

Truth be told, there really is no definite answer for this. You might forward test your EA for 8 whole months, with great results, and then the EA might wipe out your account in January and February. Or you might get great forward test for an entire year! And the EA might tank out on you the next year, due to some unexpected economic news. Naturally, one shouldn't rush from back testing to live testing, forward testing is necessary, but forward testing on a demo can never be absolute proof as to what your EA will do in the future. With that being said, and at the risk of contradicting my above statement, I would say 3 months of forward testing is generally ok, but only for the sake of understanding the EA’s “general weaknesses and strengths.” EAs must be managed, and should never be trusted through and through.

He doesn't have an EA. He tested over 400 trades. If he honestly and objectively tested, it's time to let it rip (on a small account). There is no basis for suggesting 3 more months of testing.

well, unfortunately I don't have an EA (I wish I do though, it is nice to know that some machine is always on 24-7 making money for me ), so I dont really know how objective I was during the backtest. I guess this forward test can be a way to really test the system objectively and also getting to know my system better like yoda said.

thanks a lot
cheers.

Nothing ever going to be the same you can foward test for ever but the market are always changing what if the euro in two month was back 97 or 1.60 the fact is we will never know... you just got to let it run on a micro mini and see what happens % are %. If your happy with the result and if it lines up w/ your backtesting in theory you should be in good shap


turbo