wilmott.com

Hi everybody!There are a lot of papers on trading strategy optimisation. However, it seems most if not all of them are about generation of trading signals, while the issue of money management is either untouched or simplified to finding optimal stop-loss/profit-taking points.Are there any papers on optimization of money management rules? How one could formalize money management rules in a systematic way for their further optimization with genetic algorithms/programming? I am a bit lost as to me the number of alternatives looks huge. Hints are very much appreciated!T.

.........turtletrader.com........Van Tharp / iitm.com..........Charles LeBeau's book, "Computerized Analysis of the Futures Markets". They have good coverage of the subject

ralph vince, "the mathematics of money management", introducing kelly and optimal-f. need to be careful about it IMHO.balsara (i think that was the name), he wrote something, but i fi remember correctly it was pretty poor.elder, i thnk that was the name of another guy. the whole subject is important but hardly covered in detail.

There is a new book just published about January, 2004:QUANTITATIVE TRADING STRATEGIES by Lars Kestner, V.P. Equity Derivatives Trading, Citibank Corp. & Invest. BankChapter 11 NEW TECHNIQUES IN MONEY MANAGEMENT(included the follow sections, amongst others)THE KELLY CRITERIA [ . . . I don't recall, but Kestner may have just been summarizing Kelly rather than advocating it . . . ]VINCES OPTIMAL F (pg. 316) Ralph Vince was the first to apply Kelly's work to trading (1990). Vince calculates the optimal leverage point using trade-by-trade returns. His version of "optimal f" is solved by maximizing the ending wealth of a series of trades by betting a fixed percentage of the largest losing trade. The idea is that the largest losing trade will occur at some point, so we risk some fraction of the largest loss, which maximizes ending wealth. Although Vince's method should yield beneficial results, using only daily return data from a strategy will provide a tighter, better measure of optimal leverage due to increased number of data points. Vince's method suffers if the number of trades is small enough that the ending wealth curve becomes jagged. When the ending wealth curve is not smooth, it becomes difficult to determine the true optimal leverage point.AN IMPROVED METHOD FOR CALCULATING OPTIMAL LEVERAGE     I believe the best method for calculating optimal leverage involves maximizing the median ending wealth. Harry Markowitz (1959) showed that maximizing median wealth is equivalent to maximizing the mean logarithmic return . . .Does the above explain why a number of Wilmotters have commented that they think "Optimal f" is a bit questionable?

i do not use kelly. i am afraid if i run twice on the same time series, first to derive signals and simple results per trade, second to derive the optimal betSize by using the result of the first run. i still do not know what to think about it. i heard that ralph vince tried it as a cta himself and failed. but i do not know details and background.peace

The Kelly formula is only valid for a distribution of possible returns that has a finite number of outcomes (card games, et cetera). Further, it can be shown to be implied by Vince's f - formula.Personally, I don't find anything wrong with Vince's work ( other than sloppy editing - typos and bad grammar). The problem is the idea of how you're going to model your distribution of outcomes. Statistical inference in general is where trading delusions come from.