Hi all.

I've been trying to test the attached EA and keep on getting incredibly bizarre results (you'll know if you test it). They change with every broker and it seems that it may all be messed up because of MT4 build 200. The results I get are in some cases good profit, other cases, total loss (wipes out account).

Could you help out in testing it and posting the backtests for the year 2006 here please? I've been trying all day to get it tested properly but couldn't. I'll have to do some forward testing as soon as I get some reasonably accurate backtest results. The ea is called superscalper. It's supposed to be tested (as most scalpers) in 1M timeframe and for EURUSD only. Also, make sure you test only for 2006 only (not earlier years). You can also use the default parameter of risk or increase it [it's integer value, no limit].

I warn you though. It's a superscalper, so you may end up with too many orders (tens of thousands?)

Thanks for your help..

i_me

Attached Files

SuperScalper-test.ex4 (12.4 KB, 60 views)

Just my opinion...

Many EAs will give very different results with different brokers or because of a new backtester because they are too much "pip sensitive" and then instable.

For instance it can easily happen when an EA has been optimised and that a change of a few pips in the parameters (TP or SL for instance) will make a big difference in the final result. That shows instability of the strategy.
And as different brokers have different prices (few pips but for an over optimized EA it's a big problem) it is the same problem. Same thing for a new backtester as it will modify the behaviour inside the 1M candle.

In my opinion, you should try to get similar results for different brokers or for a close range of parameters (of course results will be different but at least they should be in the same order of magnitude), and then that will show that your EA is robust to small changes.

small pip trade EA will respond *much* differently on different data. There is so much variation between the brokers it isn't funny. We have been fighting this problem on the Phoenix thread. We are going to try and move toward one broker for testing that has decent data.

I agree with your analyse, but I think that the variation is a good test for an EA. If it can't resist to variations of something like 5 pips between brokers (I guess it's usually not much more than that) how can we expect that it will stand well for futures unknown data?

I agree, and we mostly solved this in Phoenix by waiting for consecutive positive trading signals instead of trading on every signal that comes through. That reduces strange EA behavior from broker price manipulation.

Thank you for this hint. I'll see how it works out!