London
Experience: Intermediate
Platform: Ninjatrader
Posts: 2 since Nov 2010
Thanks Given: 3
Thanks Received: 1
|
I initially followed the automated machine learning route to eliminate my decision making from the equation.. However you soon become evident that such models require just as many important decisions..
..number of inputs, type of inputs, amount of optimisation, back-test sample size, sample topography, when to re-optimise, when a equity curve down-turn is too much...
So as with most things the software has a learning curve and an amount of necessary screen time..
The software does make models that earn.. However my experience is that all the models will have periods of major downturn. The 'art' is identifying the downturn within time or mitigating for it with a portfolio of un-correlated models.
|