Quote:
Originally Posted by limitup  Ken:
Name one EA or automated futures trading system backtest that even came close to performing the same in real market conditions. I have traded the currency futures market way before retail spot currencies even started. My overall point is that backtest can not determine the overall effectiveness of a trading system. The reason is that fundamentals can not be predicted. We are traders here ...not programmers....this is something that has been forgotten in the MT4 world. So, in my book, a backtest can't work in the real world...it is worthless. Backtesting serves only one good purpose...and that is to see if code is working properly for development. 9 times out of 10 a trading a sytem based on historical data will leave your wallet empty. We are here to make money...not digits. If you use robots based on historical data...one day you will be caught on the wrong side of a lock limit market. Markets are ever changing...and that will never changed. This site promotes the only way effectively to evauluate a EA is on a live acount. I suggest you take that advise. Demo is a decent playground...and even there bots that claim AI, loose.
L |
Backtesting works fine on many EA's. Megadroid tests very similar to its real running. Its one of the best ways to test different parameters ans see how they would have worked on past data, and many times it does determine the overall efectiveness of the EA.
Further it is the best and the quickest way to test and see how a new EA runs. This is not a performance test to predict the future, but a style test to see how the EA trades, how it enters and exits trades, how long its usual holding time is, whether it trades in a discernable pattern, whether it trades in a set time period, whether it progresively adjusts its stops and manages a trade, or just puts the trade out there with a stop and target.
You may not like backtesting for yourself, but the limitations you claim are pretty lame. Backtesting is the natural percurser to real time testing. Through a good backtesting 90% of commercial EA's can be eliminated immediately, and save you from wasting your time. Some fail right off because of the ignorant simplicity and limitations in their programing. Its fairly easy to spot an EA that doesnt manage its trades, or that simply runs off an indicator. Others fail simply because they dont fit ones trading style and needs.
L, your attitude is quite condescending and it doesnt really sound like you know what your talking about. Markets are ever changing, yet the patterns constantly repeat. EA's dont trade on fundamentals, they trade on mechanical systems, and these can be tested. A good EA needs to react to and filter price action, and this ability, or lack of it, is obvious in backtesting. Through backtesting its easy to see a simple EA that just runs off an indicator and takes every trade that come in. Its also easy to apply different filters when they are available and see how they affect the trading. This dosent promise perfection but its miles ahead of just putting an EA on an account and running the default program for months.
Backtesting is essential to optimization and systems design, and it is the necessary first step, before forward testing. Otherwise you waste a lot of time and never come near the optimum understanding and running of the program.