Sep

19

Optimal PortfolioRalph Vince, in Portfolio Management Formulas, says money management analysis is as important as trade system selection. Daily Spec's own Dr. Phil McDonnell improved Vince's management ideas by using return rather than drawdown in his soon to be published book Optimal Portfolio Modeling, which I look forward to buying.

Vince mentions a number of staking schemes. Vince and Bachelier refer to the basic martingale. The martingale word is derived from an equestrian device attached to the horse's bridle to prevent the horse's head from rearing back, but allowing it to lower forward. A martingale staking scheme doubles down on every loss and takes profit on the first win. The problem with the martingale is that when there are too many random consecutive losses, capital exhaustion, or casino or exchange limits, may stop it. The small martingale adds bets down per a predetermined list. Stopping is another staking system in itself. The anti-martingale is known as pyramiding. Fixed fractional system is a small anti-martingale. The reserve strategy trades base plus a reserve of the winnings. A fixed stake is another system. Vince says a proportional fixed fraction is the best. The method to determine the fraction or multiple is the question. Taking money out of trading is another staking method. Adding down and selling a portion of gains, lowering basis but keeping stake constant, is another staking system.

Some years ago Dr. McDonnell proposed the following management formula, and has further explained it to me:

Maximize: C = Sum( f * ln r(i) ) / n

 maximized over all f.

where:

 n - the number of outcomes in the backtesting
 f - the maximum fraction of your capital to invest
 r(i) - relative profit outcome for backtest case i,
 note a 1% profit means r(i) = 1.01, a 2% loss
 means r(i) = .98
 ln is the natural log

Determining the stake on an additinoal forward expectations factor would improve return as well. The goal is maximum captial appreciation and looking at computations and equity growth is as important as trade selection as long as the expectation is positive. A losing system is always a loser and no money management scheme can turn it positive.

Vince bases his calculations on a single trading system where the consecutive losses are independent. He says the calculations are harder for multiple systems, however expectations for multiple winning systems and the equity curve should be better and smoother, which should affect staking.


Comments

Name

Email

Website

Speak your mind

1 Comment so far

  1. gabe on September 19, 2007 9:33 pm

    The main problem with trading at the optimal f (maximizes the growth function G(f)=P*ln(1+B*f)+(1-P)*ln(1-f); P prob of winning;B ratio Win/Loss) is that one needs to estimate precisely the distribution of outcomes of trades for the system to work, and I don’t know if it can be done, even if one trades a fixed program (suicide?).

    The reason for this is because the system achieves P&L exponential growth in the asymptotic sense, and due to the arcsine laws effect - to quote Vince himself - “the time of the longest drawdown (not necessarily the worst, or deepest, drawdown) takes to elapse is usually 35 to 55% of the total time
    you are looking at.”

    So basically you must be willing to be underwater 55% of the time you trade… a FIXED SYSTEM.

    That being said, I found his other book “The Mathematics of Money Management” very useful and I recommend it(can email it if interrested)

Archives

Resources & Links

Search