It's the worst performer by far on acidburn's Demo, which is hard to understand as it's supposed to be the combination of Z1 and Z2 - the big winner - which itself isn't even at breakeven...
Definitely Z12 is not performing well. But, if we normalize by the required capital as reported by the backtest, even that is an understatement. With current 100/20 margin/risk setting, the backtest reports 7921 as required. And Z12 is currently (50000-46427) = 3573 underwater. That translates to a 45% drawdown which is totally not acceptable in my book.
I'm close to rule some or all of the Z's unprofitable and terminate the demo. If I decide to be selective, Z12 will obviously be the first one to go.
Secondly, Acidburn's demo has been running for six months now. It is still in drawdown. I am pretty sure there is no six month period in the back test with a drawdown. The Zorro people tell me that it is probably because the demo is started and stopped so much and there is some learning that is lost. This indicates to me that it is too sensitive to this 'learning' and that makes me nervous.
The truth is that recently strategies in the demo have had long (read: multi-week) periods of uninterrupted running. Even when they were restarted in most cases the restarts took place on weekend, i.e. when the market is closed. As the phantom trading has been turned off since the beginning (and only that component should be somewhat sensitive to restarts) I'm just not buying the explanation that restarts could have been the reason for the bad performance. And I knew that restarts would come into spotlight sooner or later, but I really don't believe one bit that running uninterrupted for months would have changed anything. Not to mention that with such approach I would probably still be running 1.14 or so...
