Your worst EAs (post them)

Hi,

recently I heard of an approach, which can deliver best -opposite- results.
If you have a really bad EA, which performs negatively (min. 90% backtest results/or live demo trading), you have to change settings to the opposite, et voila... you'll receive a very good EA.

Does anyone of you has a very bad EA - and is willing to improve it?!
Post them here {with mql4 code}

This topic has been discussed many times before. Here are some reasons why reversing a losing system won't necessarily create a winning one:

#1. You can't reverse transaction costs or slippage, which always work against you, making all of your wins smaller and losses bigger. These have a much greater effect over the long term than many traders realise, especially if your EA is trading small moves (intraday, scalping).

#2. If your broker is trading against you, or otherwise cheating you in some way that you're possibly unaware of (watch this video), he will continue to do this even if you reverse your system.

#3. Any market reactions that worked against you (e.g. stoploss hunting) will continue to do so, even if you reverse the system.

#4. Leverage decay will work against you, even if you reverse the system. For example, if you lose 20% of your $1,000 account, you now need 25% return on the remaining $800 just to return to breakeven.

#5. Robust systems tend to exploit genuine market inefficiencies. It's just as difficult to create a genuinely losing system as a winning one, because you would still need to know what these inefficiencies were, to trade against them. Most systems are impotent, rather than losing; in other words, they would return to breakeven if traded for many years (except that they lose money for reasons #1 thru #4).

#6. Following on from point #5, it's easy to be fooled by randomness. For example, look at the equity curves generated by 8 different EAs in the image below. ALL of these have 1:1 RR, and a win rate of 50%, hence (everything else being equal) all of them will eventually return to breakeven (the black line) over a large enough number of trades. But after only 500 trades, EA #1 looks like it's performing much better than #8, when in reality, it has just been more 'lucky' to encounter favorable market conditions thus far. This demonstrates why 500 trades is nowhere near enough to exhaustively test a system.
#7. Markets can change character over time, due to factors like changes in belief systems, economic shifts, and advancing technology (for example there are a lot more HFT algos running today then there were a few years ago). Hence systems capable of beating the market a few years ago might not be as effective today, and the same applies going forward into the future.

#8. Finally, some systems cannot be exactly reversed; for example a trailing SL cannot be reversed; nor a system whose position sizes are dependent on the results of recent trades.

If long term success was as easy as reversing a losing system, everybody would do it and we would all be profitable. But then it would have to be at somebody else's expense, as trading is a (pseudo)zero sum game. The bottom line is that your EA is competing against real-life experts who are just as keen to win as you are, and they can potentially change the rules, which your EA is not smart enough to recognize, nor flexible enough to adapt itself.

FWIW, I posted some more thoughts on EAs here.

hanover, you are, as always, a total Pro

Just one question: Is there a possibility to avoid the negative facts, that a retail trader (EA) has?!
spread, slippage, hunting SL (hiding SL & TP), etc.
(I heard of special relationships, for example with IG Markets, coders developing API EAs for clients)

You would need to discuss any special conditions with your individual broker. (Here is a list of MT4 brokers, just in case it is useful to you).

I'm not saying that it's impossible to successfully reverse a losing EA, just that you should probably consider the above points, in terms of determining WHY the EA is losing.