Holy grail for me is a system that have >= 80% wining trades and a max drawdown of 10-15%. Stable equity curve and is a generic strategy that always works in any kind of market, it worked fifty years ago and it do work today.
Win rate doesn't mean anything unfortunately. 40% win rate can outperform a 80% win rate system on all other metrics, profit factor is much more relevant. Drawdown is only important in relation to the returns (look at recovery factor).
I'm afraid my holy grail system is something like this...... [Wheeling trojan rabbit up to castle gates] MUTTERING GUARDS: C'est un lapin, lapin de bois. Quoi? Un cadeau. What? A present. Oh, un cadeau. Oui, oui. Hurry. What? Let's go. Oh. On y va. Bon magne. Over here... [rumble rumble squeak] ARTHUR: What happens now? BEDEVERE: Well, now, uh, Launcelot, Galahad, and I, wait until nightfall, and then leap out of the rabbit, taking the French by surprise -- not only by surprise, but totally unarmed! ARTHUR: Who leaps out? BEDEVERE: Uh, Launcelot, Galahad, and I. Uh, leap out of the rabbit, uh and uh.... ARTHUR: Oh.... BEDEVERE: Oh.... Um, l-look, if we built this large wooden badger... [twang] ALL: Run away! Run away! Run away! Run away! [splat] GUARDS: Oh, haw haw haw.
Very entertaining. Gave me a chuckle. But we still don't know if a sparrow(or two) can carry a coconut. ..... Something like my system.
you gotta love these competitions. take 1000 curve fit submissions and rank order them. voila, the top 5 must be real because, well they are the top 5! but wait a minute, why enter such a competition when the award is a negligible gain compared to actually trading those profitable algorithms???
I would assume people trading their own money aren't relying on the randomness of curve fitted submissions to "win" a World Cup.
Koch's methods aren't as robust as the Superbands systems, and I can't tell you how frustrating it is to know there's no difference between what he's doing now, and what that system has done since it has been tracked. 103 trades is not statistically significant if it were in a much shorter time period (like 4 months). There are 1400 bars of data that generated that backtest, and 1400 bars is more than enough validation to determine the profitability of the method.
I have an index system that makes 10% a year. So i test randomly on different ETF and get very different yearly results like -20% or +50%, but average is 10% still and hovers 5-15% for each etf. very weird. its like a bullish drift.
Why didn't you trade your SuperBands system then? You said it yourself, "Money. Some people got to have it". It's an honest question at this point. While a longer simulation period helps, a low trade count is never acceptable. 1400 bars is around 5 years of data, not a lot at all. By your own logic, a system that generates only a few trades during the 2008 fallout is statistically viable as long as you meet the 1400 minimum.