>Backtesting is a useful first step as if some particular system can't >be demonstrated to perform well IN sample then it's probably totally >useless Thats all the framework is required to do. It's up to the user to inspect the trading system/code further.
>But neither methodology will provide any clues as to how the system >might be expected to perform at the right edge of the chart ... I can't disagree with that. Fred wrote: > > Backtesting is a useful first step as if some particular system can't > be demonstrated to perform well IN sample then it's probably totally > useless ... An MCS evaluation may add something to this as the > liklihood of DD's expected with the same list of trades might be > signifcantly higher then one otherwise would otherwise expect ... > > But neither methodology will provide any clues as to how the system > might be expected to perform at the right edge of the chart ... > > --- In [email protected] <mailto:amibroker%40yahoogroups.com>, > "Michael.S.G." <[EMAIL PROTECTED]> > wrote: > > > > Fred, > > I was thinking a 10 year span would give a reasonable base > test. > > Then if a system looks promising, Then the user can download the > system > > and perform test on various other ranges to their hearts content. A > > system may also be tested on a completely different data set > (different > > exchange) for a general robustness test. Of course a system can be > curve > > fit, But this would quickly be noted and posted to the comments > section > > by other users. As it's an open souce project, Your more than > welcome to > > assist in resolving or creating workarounds for such issues as you > see > > fit. And if you don't see backtesting as a useful tool at all, Then > > there is no need to participate. > > > > ATB > > Michael. > > > > > > Fred wrote: > > > > > > As stated previously, back test stats only really tell one about > how > > > well one has managed to curve fit some particular system to the > data > > > that's in play. It tells one nothing about how any given system is > > > likely to perform out of sample which is of course where all > investors > > > and/or traders use real money. > > > > > > --- In [email protected] > <mailto:amibroker%40yahoogroups.com> <mailto:amibroker% > 40yahoogroups.com>, > > > "Michael.S.G." <OzFalconAB@> > > > wrote: > > > > > > > > Backtest Framework is now known as "AB Trading Framework" > > > > > > > > It's not much, But here it is anyway.... > > > > 1st Alpha release of "AB Trading Framework" is available at > > > > http://www.amibroker.org/3rdparty/AB_Trading_Framework/ > <http://www.amibroker.org/3rdparty/AB_Trading_Framework/> > > > <http://www.amibroker.org/3rdparty/AB_Trading_Framework/ > <http://www.amibroker.org/3rdparty/AB_Trading_Framework/>> > > > > > > > > > > > > > > __._
