EURUSD 1 hour chart, back-test from 1-1-2011 to 1-2-2012, wins 70% of the time, and returns 42%.
This strategy looks for "inside bars" and places pending trades above and below the high and low of that bar. It's very easy to trade manually, and I'll share my "inside bar" indicator with you here.
The robot code is not available, sorry ;) If you're familiar with programming at all, it's an extremely easy strategy to turn into code.
So with such a successful robot, why am I not retired?
Well, when we back-test on the GBPUSD, things do not look so great:
That's a $2181 loss, despite winning 64% of it's trades.
With some optimizing of the stop-loss, etc, this robot did return a small profit on the GBPUSD, but they weren't settings I was comfortable with (poor risk-reward ratio).
So what's the problem? The strategy is crap? A problem with the robot code? More dodgy history data? Problem with the back-test?
Probably a little of all of these, but it could also have something to do with the GBPUSD itself. I'm not sure what exactly, but I've noticed that very few of my robots back-test well on GBPUSD. If they do well on EURUSD, they usually make a little profit on AUDUSD, USDCAD, USDCHF, and sometimes even USDJPY, but they more often than not completely bomb on GBPUSD. They don't just lose a little bit, they lose badly.
Sometimes optimizing settings helps, but usually not.
One theory I have (despite the daily ATR values vs EURUSD being approximately the same) is that the GBPUSD moves a lot faster than the other pairs, and thus my settings should be wider - bigger stop-loss, bigger take-profit. I have no hard data, but back-tests of other robot designs seem to support this idea. Since I code robots primarily for EURUSD though, I am not completely sure. One day I might get around to investigating this further.
It would certainly make me feel more confident in my robot back-testing if I had similar results across most of the currency pairs.
