fxozgirl wrote:Gary, you are right with the rules...those are exactly right! It's the EA that is closing the trades a bit funky, not sure why, perhaps you can take a look at the code, although, I wouldn't waste too much time on it as when Steve gets time he's coding a new EA for me with fully flexible options so we can test a wider range of parameters including different time frames, stoch levels etc.
Ah. I thought the results you'd been posting were from your manual trading, not the EA.
Aside from my funky results, how were the results on the tests you've been looking at?
Lousy. That's why I haven't been pursuing this any harder.
Results from the last 12 months:
AUDCAD: -$758
AUDCHF: -$121
AUDJPY: +$201
AUDUSD: +$126
CADCHF: +$1727
CADJPY: -$1436
CHFJPY: -$444
EURAUD: +$913
EURCAD: +$2565
EURGBP: +$1453
EURJPY: -$695
EURUSD: -$1212
GBPAUD: +$927
GBPCAD: +$457
GBPCHF: +$2322
GBPJPY: +$168
GBPUSD: -$658
USDCAD: -$1352
USDCHF: +$782
USDJPY: +999
That's trading 0.1 lot, so $1 is roughly 1 pip. So trading 20 non-NZD pairs (omitting EURCHF), it got a whopping $500 per month. And that sounds better than it actually was. This is what the whole portfolio looks like for the last year+:
Gday.gif
So it made about 750 3-lot trades, and most of the year it went sideways. And this test didn't include spread, commissions, etc -- include those and it would look a lot worse.
With the results you've been reporting (+2573 pips in 30 days!) I decided I needed to look closer. But if you look at the equity curve, you'll see there was a $3000 run-up since early September. It may be that my test is reasonably accurate, and the last 30 days were just unusually good. This is why I don't trust short forward tests. I want to see how the system performs over the long term. If my test is accurate, this isn't a system I'd want to trade.
You've been trading Gday (manually and EA) a lot longer than 30 days. What kind of results have you been seeing? Is the 2500 pips in the last month typical, or unusual?
It may be possible to make this work by cherry-picking the pairs you trade. (And you may be doing that already.) Some of them worked a whole lot better than others -- e.g. the CHF pairs seem to do fairly well for some strange reason. That's especially puzzling since the CHF peg means CHF is basically a mirror of EUR, and the CHF pairs look a lot better than the equivalent EUR pairs. (!?) When I tried this pair-selection filter before, it didn't work all that well -- the "pairs that work" didn't seem very stable. I'd select the best pairs from one period and walk it forward into another period, and those pairs would stink in the new period. But I'll take another look at it. I'll run it on 2011, pick the pairs with the best Sharpe ratio (so pairs with big up-and-down swings don't get picked just because they happened to end on an up swing) and see how those pairs perform in 2012. Will let you know what I find. |
