Hi, 

In portfolio backtest of a futures symbol, I apply a simple  
antimartingale way of position sizing  by indicating  < 
positionsize= -50; >  There are about 400 historical trades. 

When looking at the tradelist of the portfolio backtest, I found that 
in the beginning one hundred trades or so, the number of the contract 
traded changes nicely according to the result of the individual 
trades plus the effect of the sizing algorithm , increasing or 
decreasing by one contract at some point co-defined by the margin 
requirement of the contract.  Thus the contract traded increased from 
5 contracts to 11 contracts. Then something strange happens:  after a 
small losing trade, according to the cum profit ,the initial equity 
and the margin requirement, the number of contracts to be 
bought/shorted in the next trade should remains at 11 contracts, but 
in the  backtest only 9 contracts are traded. And from that point 
on , the number of contracts traded in the backtest do not conform to 
the theoretical value ( which is correct bcz it goes with the sizing 
definition ) from time to time, and in the latter half of the 
tradelist , I would see adjacent 10 numbers of contracts traded 
as "27,27,29,25,13,30,17,25,26,30"   Till this point , I believe 
something must go wrong here, because the result of these ten 
individual trades should never cauze such big jumps in the number of 
contracts traded.



I want to know if some one have met this before in the portfolio 
backtest for a sample of more than a couple of hundreds trades. And 
what may be the possible cause & solution to the problem.

Thanks for any hint  and Happy new year to all !

huanyan


Reply via email to