Hey dom are you gonna keep the system going? With all due respect, it seems the performance is pretty mediocre.
If you don't mind, I'm curious: What did you learn from the drawdown? I mean, I'm sure it wasn't expected right? So, then how was it overlooked in testing or what do you think happened? But, more importantly, what did you learn?
The max drawdown I experienced was in October, slightly over -22k from July P&L peak. Even though this was above the max allowable drawdown I had defined for that system (-32k), I triggered me to re-do the statistical analysis, and get rid of number of patterns with (too) small sample size ... It is very likely that these patterns were just over-fitting the system to past data. The price patterns that I kept have 1200, 4200, 730, 2800 & 110 samples (the last one is narrow by definition, but has a performance level way better than any of the other patterns). The other lesson, was to always trade with a stop, even though the performance in backtest is lower with stop than w/o stop ... it became clear that stops are necessary to handle series of large losing trades falling outside of the backtest envelope. The performance since I turned the new version on mid-October isn't stellar (just under +10k for 1 contract, ~110 trades), but more or less in-line with my expectations for this version.
Results for the week ending January 31, 2014: - 3 wins ; 4 losses ; net +1065 The week started well, the system making a new P&L peak (measured since I changed version in October) about 2k above the prior one. The 2nd half of the week wasn't so good, the current drawdown from that peak is -1800.
no offense DOM but your best sample size is way too small, all bets are off when there is no more fed money in markets, look how different they are with only 20 k pullback (10 more starting in Feb.) you needed to have traded cl before and after fed money to know how differently the instrument trades (night and day) throw all your historical data out the window for last five years if QE ends finally! Frankly, even the bots will have to adjust once real liquidity - in terms of volatility comes back to oil markets - the intra-day dumps used to be mind numbing 4 and 5 bar absolute flushes before QE arrived.
No doubt there are way better traders than me, and way better automated trading systems than this one (or any one system of mine). That said, it has currently 548 live trades in 10.5 months, is 100% automated, always in the market (really ... either long or short, never flat), the P&L is tracked on the RAPA website. I am always eager to learn ... feel free to PM me if you want to show me a better way.
Results for the week ending February, 7, 2014: - 4 wins ; 4 losses ; +980 The system made a new P&L peak this week on Thursday (+670 vs P&L peak last week). This is one of the rare times the system is flat for the week-end (it got tricked into short this morning, and stopped during lunch).
Happy Valentine's Day! Results for the week ending February, 14th: - 5 wins ; 2 losses ; net +2085 The system made a new P&L peak this week, +1450 vs last week. We had repeated connectivity issues on Sunday night, Monday evening & night, Tuesday evening & night, Wednesday evening & night. In all cases, lost both TWS & IQfeed connectivity and I had to restart everything - sometimes multiple times in a few minutes. What is the strangest, is that at no time did I lose connectivity to my VPS. Anyways, no issues Thursday evening & overnight, which was really great as I was at the Elton John concert here in Ottawa (decent set list & show, but the acoustics of the arena is a disaster, which ruined much of the event).