5/22/2007:After looking at his performance summaries from WL his sharpe was only 3.38. My system has a backtested sharpe of 5.5. His DD when it does happen is greater than 10%. 220 trades, win percentage 80%. Losses are just as much as his profits. Up some, down another, but with the high win percentage appears good. Probably not possible to trade. Even if there's a quasi performance guarantee, 200 bucks per month prices himself out of the market.
Avg. Profit is 1.5%, avg loss is 1.3% with an expected value profit/loss of .9% approximately.
The point is that there are better systems. His annualized return at the moment is 154%, but is overleveraging himself from his backtest, boosting returns, and getting attention from his pushing of the system's position sizing.
Equity curve in the backtest was extremely suspicious, and I suspect peeking with this system. We'll have to see how it does.
I would like to point out the importance of the sharpe ratio. That is ultimately the only ratio that should be observed after a three year period. Not that 3.3 isn't high, just that other systems are much higher. When he does drawdown, it will be a significant amount, probably greater than 8%.
The highest annualized return I saw was 100% and with 18% DD. I can scale some of my other systems to get returns significantly greater than this with less risk.
If you haven't read my commentaries, I think you should.
Also, I want to point out that I am a CFA candidate, and licensed professional.
7/8/2007: I still think he's high on the subscription rate, but definitely more reasonable.
7/17/2007: This is a guess, but he's probably short right now.
8/30/2007: Still coming down from a very high level. Not very good for sharpe.
9/2/2007: Well, I always contend that when the system draws down more than your backtest that it has ceased being effective, and that would be my conclusion with any system. It's at that point.
Another observation I have is that I've found a greater prevalence of curve fits with systems with win percentages between 75 to 90%. Almost always these systems don't work that well. When they're over 90% and greater than 30 trades, they usually work going forward with those same percentages. Something about that win percentage is always suspicious, especially when it does not perform as you expected it to.
9/5/2007: Made losers this week (23)%. Man, it's the fall of rome. All the way up above 160, now barely 10% up.
9/10/2007: OMG, I can't believe that a backtest could look so good and have this at the end of it. Man, btest suggests curve fit after most recent DD. I actually thought it would work a few months ago. But now, with 40+% DD, I seriously doubt its longevity, and I would say the same with my own systems if that were to happen, but since I am very experienced at detecting fatal flaws, I know it won't.
9/10/2007: So you can get an idea what I think is wrong here, here's what I'm pondering. Say a system returns 50% annually over a 5 year period, like this system, say it wins 80% of it's trades. What I think this means is that it is a kind of spurt trading system. I want to say that I have developed systems like these with nearly identical characteristics in terms of performance and win percentage. What happens is that the system will do well over a long period of time, and then suddenly in shorter time frame than it was doing well give back a significant chunk of the system's profits. What I spoke about earlier was the significance of the win percentage. When a system wins less than 90% and greater than 3/4's of its trades, it seems to me as I would look through the trade history and observe the equity curve that it would work in what I called 'spurts.' These spurts would win the historical percentage of the time, usually in long streaks, and then the other 20% of the time give back a lot of its gains. It is this 20% loss streak where the curve fit happens. They will skew your stops and targets, or whatever other indicator you might be using and trick you into thinking it will be good going forward. That's the curve fit part, and I want to say that no vendor intentionally does this. It's just part of the developmental stage. My approach were I to have a system where this happened might be if it was a average DD, go back and recalibrate, usually by taking fewer trades and increasing the win percentage but not the APR. If it was an enormous DD, say double historical DD predictions, I'd just quit, and move on to something else.
In his case, I'd suggest quiting if there was no evidence prior to this that his position sizing and fundamental assumptions would lead to a doubling of the system's DD. If he didn't expect it, then the system should stop, as it has ceased to be effective. There, that's as simple as I can make it and address all of the possibilities. I actually think most of the problem was a position sizing flaw, not timing, though that could be improved as well by loosening the thresholds.
Ex post edit: Going on the last thought, that it was a position sizing problem, I point out what I had said in earlier posts, he's showboating and overleveraging himself. That should have been the red flag for anybody trading the system. It happens with every futures system. You just can't control market exposure well with futures, because no matter if initial margin was 4k or 2.5k, the ticks are valued at the exact same amount, usually $12.5/tick or $50/point. That's where I don't approve of futures for my own personal trading.
9/11/2007: Vendor states the DD was all caused by his own discretionary trades. Well, I gues this goes to show there's no room for non-systematic trading on this website, but vendor is probably no longer serious about offering subscriptions and claims his system has done well and not what C2 has shown.
10/22/2007: He has not traded the last several trades I've received in confirming his results. I guess it's a lost cause now the results on C2, but his system probably does work, he's just not following his discipline on C2.
2/12/2008: I had been receiving e-mail alerts for this system. For six months he showed a profit per contract of 61 points. In a matter of a few days, lost close to 500 points per contract on 4 contracts. Vendor thought it was time to go back to the drawing board.