from personal experience, having run many many walkforward tests, most systems can be optimized (curve-fit) to appear to give good results using in sample data. But ... results using out of sample data are usually poor. So, you use IO to do walkforward testing and determine if your trading system will really work. 98% don't! Larry
--- In [email protected], "murthysuresh" <[EMAIL PROTECTED]> wrote: > > still trying to get it into my head. thanks for your points. > > --- In [email protected], Keith McCombs <kmccombs@> wrote: > > > > Does your "simple trading Practice" use any market indicators? > > If so, are the market indicators a function of any numeric > parameters, > > like smoothing period for example? > > If so, have you ever wondered if maybe there better values for the > > parameters? > > If so, did you try backtesting different values? > > If so, were more than two parameters that you wanted to test? > > If so, did you try it? > > If so, did it take all day and night to try all the combinations? > > If so, thats why you should try IO. > > If not -- don't bother. > > > > murthysuresh wrote: > > > > > > i have always wondered, why would someone use the IO. there is a > AB > > > internal one and a Fred external one. i am trying to understand > how it > > > will be useful. > > > my approach is a simple trading practise. doing woodies zlr in a > > > trending market. but i always wonder how IO would be useful to my > > > analysis. > > > > > > > > >
