I've noticed this effect also, and never quite figured out why the equity curve was different than the raw numbers for a WF run.
I've always gone with the lowest number, to err on the side of conservatisim. But it would be nice to know why the difference occurs. --- In [email protected], Howard B <howardba...@...> wrote: > > Hi Jack -- > > Can you provide more information? Code being used, setup for walk forward, > etc. Send me a private message, if you wish. howard at blueowlpress dot > com. > > Thanks, > Howard > > > On Wed, Jul 15, 2009 at 12:24 PM, jacklweinberg <jacklweinb...@...>wrote: > > > > > > > I seem to have a major discrepancy in the AB walk forward analysis. > > 1. The (partial) results of a walk forwrd analysis: > > Mode Begin End No. Net Profit > > OOS 1/1/2004 12/31/2004 1 15127.33 > > OOS 1/1/2005 12/31/2005 1 20248.16 > > OOS 1/1/2006 12/31/2006 1 33777.22 > > OOS 1/1/2007 12/31/2007 1 10681.52 > > OOS 1/1/2008 12/31/2008 1 18440.27 > > > > This shows that the out of sample Net Profit is: $98274.5 > > (adding up the Net Profit column for the OOS mode). > > > > However, in the ~~~OSEQUITY indicator, the total net profit is $118594 > > a difference of $20,320! very significant! Which one do I believe? > > > > This happens on EVERY walk Forward Analysis - > > Is there something wrong? Why such a difference? > > > > NOTE: ~~~OSEQUITY is over the same period, i.e. 1/1/2004 - 12/31/2008. > > > > > > >
