I've implemented a non-hedging version in Tradestation. Instead of going long and going short, it goes long and then closes the long. That has the same net effect as the two hedged positions, without hedging.
Here for example is the behavior on EURUSD on 1 Dec:
It hits long1 (A) and goes long 1 mini-lot. Then the market reverses to short1 (B). Instead of opening 2 shorts, it closes the long and opens 1 short. Then it shorts 2 more at short2, so it is now short 3.
Now there are two pending orders at long1 (C) so it closes 2 shorts, still short 1. It hits 3 long orders at long2, closing the remaining short and going long 2. It adds another 3 longs at long3 (D), then proceeds to the TP, carrying 5 longs. By comparison, the original definition carried 9 longs and 4 shorts to the TP.
My results match Dietmar's closely, but not exactly. I need to check into a discrepancy in starting times. For example on 11/30 the market moved down sharply just before Dietmar opened his trap -- but according to my charts, an earlier bar was 00:00 local (US Mountain Standard Time) == 07:00 GMT:
So I have to work out a few details like that, but other than that I think I have it working properly. Dietmar, any suggestions?
Results: on EURUSD, with 15-pip bands, starting the trap at 00:00 MST == 07:00 GMT, it has made 4000 pips in the last 3 months! BUT it went up and down for 8 months before that, and lost over 5000 pips in the last half of December 2010 and January 2011. Ouch!!!
GBPUSD with 01:00 == 08:00 GMT starting time was a bit rough too:
The system is extremely sensitive to starting time, as you'd expect. What disturbs me more is how "fragile" it is -- the good spots are very spiky, with bad values right next to good ones:
(But actually if you look closer, any time from 08:20 to 09:30 works well.) By far the best result is with a 09:00 GMT start time. I don't like the narrowness of the "spike" at 09:00, but it performs well all the way back to late 2009.
I didn't include 3 pips spread costs like Dietmar did. In my experience trading FX at Tradestation, I only see about 0.6 pip slippage on stops. I didn't bother including it here. However the costs are an important consideration. Even though this non-hedging approach has lower costs (open long + short long = 1 trade; open long + open short + close both = 2 trades), the average trade size is still very small.
I've looked a little at different band widths (fixed and ADR-scaled) and a few other things, with no huge discoveries. When I get a chance to work on this again, I want to see if I can find some exit strategy to bail when a big loss is building. I don't know if I can do that without ruining the performance on winning days.
I also want to explore other pairs. None of them are stellar performers all the time, though some (like EURUSD and EURCHF) do mighty well. But trading more than 2 of them might produce better results.
For now, the way Dietmar trades it works pretty well, at least as long as the pairs keep behaving. (Though I would change EURUSD to start at 09:00 GMT instead of 07:00. GBPUSD is good at 08:00.) Combining the two pairs helps to smooth things out a bit. Here's the result of trading EURUSD and GBPUSD using Dietmar's parameters:
It was a rough go in the first part of the year, but it's won over 8000 pips since mid March. If I can figure out how to filter out losing periods like like January, this will be a pretty potent tool.
Gary
It's true, yesterday it had missed both, due to my intervention, lol....