Well, this post is from quite a while ago ... but so far, my results have shown that this concept seems to work well. I don't even bother trying to figure out the market condition though. I just find a lot of uncorrelated (both mathematically, and fundamentally) systems that work reasonably well (no stellar, shoot for the moon results) and run them all at once. Really helps smooth out the equity curve, and I don't have to worry about corner cases in each system...
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
The one thing I never understood...
- Thread starter Corey
- Start date
Well, this post is from quite a while ago ... but so far, my results have shown that this concept seems to work well. I don't even bother trying to figure out the market condition though. I just find a lot of uncorrelated (both mathematically, and fundamentally) systems that work reasonably well (no stellar, shoot for the moon results) and run them all at once. Really helps smooth out the equity curve, and I don't have to worry about corner cases in each system...
Quote from Corey:
Well, this post is from quite a while ago ... but so far, my results have shown that this concept seems to work well. I don't even bother trying to figure out the market condition though. I just find a lot of uncorrelated (both mathematically, and fundamentally) systems that work reasonably well (no stellar, shoot for the moon results) and run them all at once. Really helps smooth out the equity curve, and I don't have to worry about corner cases in each system...
your on the right track. Chasing regimes aint the way to do it
Quote from Corey:
Well, this post is from quite a while ago ... but so far, my results have shown that this concept seems to work well. I don't even bother trying to figure out the market condition though. I just find a lot of uncorrelated (both mathematically, and fundamentally) systems that work reasonably well (no stellar, shoot for the moon results) and run them all at once. Really helps smooth out the equity curve, and I don't have to worry about corner cases in each system...
I agree. It's super hard to actually identify the different market conditions.
My list of market conditions is because it helps me build uncorrelated strategies that capitalize on different market conditions--and then run them all at the same time.
For example if you build a trend strategy, it will work well in the trends but either lose or have a flat equity curve when non trending.
So in order to get non-correlating strategies, it seems you need one that focuses on profiting in a different market condition.
Another way, I use to get more non correlating strategies is to use a different "fractal" or time frame.
Wayne
Quote from pneuma:
Add multiple markets and time frames and you will come up with many more.
pneuma
Okay. That makes sense. I do the multi time frame thing already so that does multiple it out some more. Thanks for clarifying!
I'm curious what software you use for that? I couldn't find any off-the-shelf that does multiple strategies on a single symbol easily (so wrote tz). Did you write it yourself?
Wayne
Quote from tickzoom:
I agree. It's super hard to actually identify the different market conditions.
My list of market conditions is because it helps me build uncorrelated strategies that capitalize on different market conditions--and then run them all at the same time.
For example if you build a trend strategy, it will work well in the trends but either lose or have a flat equity curve when non trending.
So in order to get non-correlating strategies, it seems you need one that focuses on profiting in a different market condition.
Another way, I use to get more non correlating strategies is to use a different "fractal" or time frame.
Wayne
Off the cuff, you can normally identify diversity through:
1) Asset Class
2) Strategy
3) Time frame
4) Geography
For example, a short-term reversion to the mean strategy running on US futures would be non correlated with mid-term break-out of chinese equities would be non correlated with long-term trend following of the british pound...
Quote from Corey:
Well, this post is from quite a while ago ... but so far, my results have shown that this concept seems to work well. I don't even bother trying to figure out the market condition though. I just find a lot of uncorrelated (both mathematically, and fundamentally) systems that work reasonably well (no stellar, shoot for the moon results) and run them all at once. Really helps smooth out the equity curve, and I don't have to worry about corner cases in each system...
this is one of the most useful posts i've seen on this forum. for all of you looking for that "holy grail", corey seems to have found it... trade multiple uncorrelated strategies on multiple markets with conservative money management. try it out and you'll see how quickly your account compounds with very little drawdown. so what if a couple of your strategies have no edge, as long as they are effectively uncorrelated to your profitable strategies they'll help minimize drawdowns and improve the combined results. by adding more strategies/markets to further minimize drawdowns, you can effectively crank up the leverage to yield some mouth-watering returns (subject to the usual capital constraints of course).
as another tip, i'll point out that you can intentionally add an unprofitable strategy to your system to improve the combined results. test everything.