What about using the Spearman Rank correlation? - Looks like providing a bit less information, but worth considering...
[[What about using the Spearman Rank correlation? - Looks like providing a bit less information, but worth considering...]] Yes, that was my experience too -- but, (and contradictorily, if that is a word) the robustness of the Spearman r seemed to be a nice quality. It seemed to have the character of amending itself toward what the future might throw at us, more than the classic r.....and that might be just my own imaginings too.
A little bit more robustness at the cost of losing a very important piece of information: severity of change in addition to the joint probability of change. That makes it very tricky to use. Will require some sort of additional model for getting the magnitude-of-change measure. All these: betas, spearman, correlations, etc, are variations on the same theme. If you are concerned about things that they don't describe, going from one to the other won't help you. The deepest problem of portfolio risk management: people tend not to know what the question they actually want answers to is. Is it the severity of known losses, unknown losses, gap risk, leverage capacity, etc?
Exactly -- they do NOT know what they want. Not just on the downside (Is it variance? Is it drawwdown? What is it?) but on the upside as well (geometric growth optimality? Inreased average return?) For answers to these very things, I turn to economic theory -- and jsut what makes people tick. (because, absent that, ...all we really know is that they tick!)
Greetings Mr. Vince... I read your books way back when the CME library was open to the public.. Your quote above begs a follow-up... Scale Trading, Geometric adding to positions, Options spreads, Grail Thanks, Lakeside
Its simple -- but I won't get too specific as I don't owe anyone any more than that. However, my criteria has been to maximize being profitable, as opposed to maximizing profit itself.
... be careful there; It's great that you've been doing well with your purely dynamic sizing based strategy; Just want to offer the observation that CPDO type of dynamic strategies did GREAT for a year or two until Dec of 2007, where they lost billions. CPPI type and a bunch of other such strategies all had similiar experience. When it comes down to it, there are only two basic "primary" form of dynamic sizing strategies: convex to market and concave to market. One will do will in trending, the other will do well in range bound. Just make sure your have properly analyzed your strategy as opposed to just basically being long one type of beta that's been doing great in the current market environment.
Dynamic sizing based strategy on equities? Can anyone enlighten me with just the general concept how it can maximize winning %?
This is a pretty classic strategy: http://en.wikipedia.org/wiki/Constant_proportion_portfolio_insurance works great in trending markets. Will murder you when the trend reverses.
An interesting wikipedia article -- but it is not what I am doing here. My point is I am using the LSP model in a manner not to maximize profits, but to maximize the probability of being profitable at a given horizon in time. This was an interesting discussion guys. Thanks.