1. win rate too high? Not exactly. 60%+ system is not bad at all. And there are short-term systems that holds a > 50% that does consistantly for the ES and SP, for at least 10 years, I think it had a Sharpe of 1.7 2. not enough data? Yes, this is the biggest problem. 3 years of data is too small to make any conclusions. 3. too many trades? Ummm... interconnected to 1. 4. WL Ratio? Ummm... 1. + 3. = 4. OK, to the point. System Development, just like a discretionary trade shows the character of the person as a trader. Oddi, obviously, has found a style of character which he wants to see in a system... which fits his trading tolerability. The best way to find a system that works is to execute it. You might get a lot of hesitation at first but just put those trades in. You'll build a iron-glad discipline to follow them which makes it easier to trade and develop multiple strategy systems to make a good diversified portfolio of systems. Regardless of what people think, a system trader, first has to develop a strong discipline to follow the system they decide to trade, no matter what. What most people do is they develop a system that makes consistant money (which is hard ) and also, to fit their style which they have no idea about. The latter not needed. Until you can say, "I can trade any system of any kind without any personal conflict." , there's no point in developing any system because you'll be dealing with unneeded bias, rather than what you really have to deal with.
So a simple answer... 1. Try to get as much data as you can. I prefer 20+ years but maybe 10 years with intraday, is probably good too. In another words, more is better. 2. Look at the Sharp Ratio. Sortino, etc. etc. 3. Unfilter all the setups. Is the system profitable just with the entry criteria? Then you've got something. Bad entries can be turned good through various setups but I don't deal with them, personally. 4. Fade. For example, there's 20 day breakouts (Turtle Entry) but this can also be a resistance ( Turtle Soup Entry ) . Develop systems for both separately. This helps you draw some valuable conclusion of the system you are developing. Well, simply just be skeptic with what you develop.
Thanks for all the info. The "pattern" i have is well - usually, call direction of market. I have been trying many "variations" of it, with many different results. The hardest part, I find, is exiting. I am trying another method of this way, which does about 100 trades a year. It seems more conservative, and I dont get killed in 01 like I did before. As for data, my reach only goes back to 99 - I bought minute data from ANF futures. I guess I will need to find more data elsewhere.
Good point you made. Win loss ratio too high because > 50% means data is curve fit to the phase the sample data is representing. When the market rotates( which I am quite sure you know about, W.D. ), amiga is a dead duck. Ten years is a market rotation point, along with 5, 30, and 3 years, I think. Of course this has been only my backtesting experience. Too many trades because of commisions. The more s/he trades, the less efficient the system. We agree on data. Based on my personality I reject all systems with win loss ratio < 3. Speaking of rotations, the wheel says 11/30 is a turning point, based on all time high only. So Monday might be fun to watch. Regards Oddi
> 50% win rate does not mean curve fit. Ever had a system that takes smaller wins and larger losses? Well it means a very high win rate. Does that some how equate to curve fit? If it does I must be in the dark.
Toyland, thank you... You seemed to have read my post... Oddi, here's a good example of what human perception does. You only perceive and grasp what you want to... Chart reading is the same thing.... Human perception and perspective defies the truth... what is important is to tune your mind towards an objective. ... Don't you love trading?
I certainly do, and the small win, large loss argument has been brought before. You did see that I reject systems with win loss ratio < 3 didn't you? I use mech systems for the very reason that human perception makes patterns pretty much useless, especially elliot and candlestick pattern that play off of wide or narrow range candles. Trading is a fabulous game Regards Oddi
Hi toyland, yes I have seen these strange systems. I would not play a system that has a negative efficiency rating, nor a win loss ratio < 3, a system with small win/large loss will have both of those undesirables Regards Oddi