Personally I would get additional data (perhaps two years worth). I had an automated system that I was backtesting over nine months of data, was getting great results, and then I added the four-month period before and twelve-month period after my original nine months of data. As I was developing the model I was confident that I wasn't over-optimizing, but adding almost 200 percent more data to the series yielded only 70 percent more trades than the 9-month period alone, and the total profit for the entire 25 months was actually less than the 9-month period alone. Needless to say, I was very disappointed; however, I'm glad I learned my lesson by giving away my cash to the data vendor rather than the market (at least I have the data for future projects). Please keep us informed of your progress. Regards,
Jangofolly, thank you for your suggestion. Can you please give me the website of your data vendor. Thanks
P.S. I don't know what kind of data you're using for your backtesting, but the data vendor I use has tick data with optional best bid/ask price and size. Knowing that my simulation is buying the ask and selling to the bid takes some of the guesswork (but not all) out of the slippage consideration. PM me if you're interested in the vendor's info (I have no vested interest in them); I checked their site and they have CME FX data going back to 14 March 2004. I just saw your last reply, I'll send it to you...
it's very different, you will find yourself wanting to break the rules more often. my advice is: 1- 4 months is enough to start trading real money 2- trade only one contract for a few months 3- concentrate only on follwing the rules