EA performing very differently

Hi,

I'm developing my first proper EA with order management etc.
How is it, that the EA is performing so differently on different currency pairs?
E.g. on EURUSD, I can back test it for a year and have profit factor of 1.34 and 70% of winning trades and a maximal drawdown of 513 (in an account of 10k), thus making a good profit (of course there's still a lot to change and test, especially the losing trades)

But on EURGBP it has a profit factor of 0.6, losing a lot of money.

Do/should profitable EAs work equally well on all currencies or is my scenario above normal? http://www.fxfisherman.com/forums/im...milies/huh.gif

every pair has its own characteristics. There's also question of spread difference and how its being handled.

since you say this is your first EA, perhaps there are some logical errors that reflects clearly on pairs with wider spreads?

Thanks Zen_Leow. It really seems they do have different characteristics.

A follow up question: If you develop an EA for a specific pair, how do you avoid the problem of just adapting your EA and strategy to perform as good as possible on the historical data? Isn't this over-optimisation on the wrong end,
or is this how people do it:
- program initial strategy which has been more or less proven by manual trading
- adapt strategy/order/money/risk management on historical/real market
- profit!

?

for backtesting - use general (not specific settings) - this usually means that the very best results in backtesting usually arent the ones to use. test longer periods. does the EA work only in the past month or all of last year? - Use backtesting as a settings guide. if you pick the best result for each variable then you just curve fitted it.