Forward optimization

The best optimization strategy is like the one used in NINJA, I am not sure the name I do not remember, something like "Step Forward".

This is the ONLY optimization strategy that gives you some relevant expectation about what can happen in the close future. Any other strategy will give you only overfittings you have no idea if they will repeat next day.

Let me try to describe it briefly using an example.

On historical data you go back some years and optimize a set of any 24 months. Then you get the best opt and test it on the next 3 months (after the 24). Then you go back, go forward 1 month (from the 24 begining) let's say (it could be one day or any period of time), and repeat the process: optimize the next 24 months and test the best opt for the next 3 months after the 24. Just repeat this up to present date (minus 3 months) and save the results of each best opt run. You will get a very good overview if your cbot hits the future and deserves a shot in the real time future.

This methodology of optimization is a must if you want to do the best plataform for cbots trading.

Thanks, Lavio