Is this really a big deal? For the system to lose that much it will be a pretty rare event. extra trading costs weigthed the probability seems to be a minor factor
Depends on what your half kelly is. Suppose you think that your SR is 1.0 implying annualised vol of 100% full kelly, 50% half kelly. If you run at 50% vol on a quarter of your full capital, then on an average day you'd lose 50% / 16 ~ 3%. It's not infeasible you'd lose 10% in one day and have to cut positions 10%. Doesn't sound much but that can easily rack up serious trading costs. If you run at 50% x 1/4 = 12.5% vol on your full capital then an average day would be less than 1%, and your regearing trades would almost be lost amongst your other trading. GAT
trade for a few years to get good stats on your ratio then apply kelly for optimum growth. Always adjust the ratio as you gather more data thru your trading. a trader starting out using kelly is ridonkculous! and probably a lot of wishful thinking.
brainyforex.com/position-sizing-methods. Hope this help. I use the % Capital method (since I become aware of risk...) which I call %Risk: For Forex accounts: Lots to trade = Equity * Risk% / (Stop Loss in Pips * Pip Value) / 100 I don't care about margin since I the risk that I take is between 1.5 and 3%. I still use this formula for Forex, Equities and Futures. The risk that I take is related to a specific strategy (expectancy, peak of DD and duration of DD). I have 5 strategies. I take 5 different amount of risk.
Here is my current recommendation for a real-world trading fraction: Trading fraction = min[ 0.9/max[ .0001 , 2*max[ -Ri ]_i=1toN ] , ½*NKF ] , where NKF = the new Kelly formula (see below) , Ri = the return (fractional) of the i'th trade , N = the number of trades. NKF = max[ 0, s1 ]*( s2*s2 - s1*s3 )/(s2*s2*s2 + s1*s1*s4 - 2*s1*s2*s3) , where S1 = sum[ Ri ]_i=1toN , S2 = sum[ Ri*Ri ]_i=1toN , S3 = sum[ Ri*Ri*Ri ]_i=1toN , S4 = sum[ Ri*Ri*Ri*Ri ]_i=1toN.