Great explanation, thx! El 08/04/2011, a las 19:16, nonlinear <[email protected]> escribió:
> > > Let's imagine I want to create a LongDefender strategy for YM. Obviously ES > parameters won't work for this contract, so how do you set your parameters > ranges? > > Do you optimize it with wide ranges (taking a loong time) and then restrict > them by checking best performances? > > > 1. I almost always optimize for best Performance Index. Resist the temptation > to look for the highest P&L and profit factors. > 2. Optimizer takes advantage of all the CPUs on your machine. The very > powerful i7 chips are inexpensive these days. > 3. The optimization period is important. On the one hand, you want to include > as much of the history as there is, but on the other hand, you don't want the > optimizer to "learn" from the distant past. > 4. The min number of trades is important, too. If you set it too low, it's > highly likely that you will find strategies with astronomical profit factors, > but the statistical significance of those will be low. But if you set the > number of trades to be too high, you'll be forcing the optimizer to take the > trades. > 5. The divide-and-conquer optimizer is fast. Notice that you can control its > speed vs accuracy (click the "Advanced" button) > 6. After the D&C run is complete, you may want to inspect the optimization > maps and then run the brute force optimizer on a smaller sub-space of > parameters. I usually run my brute force optimization runs overnight, as they > may go through millions of strategies. > 7. When choosing the final values of parameters, I'd suggest finding the > broad plateaus from the optimizations maps, i.e. the large areas of high > performance, as opposed to isolated peaks. If there are several of those > plateaus, you may want to define one strategy for each of those, and trade > multiple strategies. > 8. After you decided to settle on a specific set of parameters, don't forget > to hard-code them into your strategy and recompile. > 9. Once in while (let's say every month or so), re-run optimization to > incorporate the latest data and to see if your best parameter set should be > adjusted. > 10. When trading your final selection of strategies, be patient and do not > interfere with the strategies by taking manual trades. > > > -- > You received this message because you are subscribed to the Google Groups > "JBookTrader" group. > To post to this group, send email to [email protected]. > To unsubscribe from this group, send email to > [email protected]. > For more options, visit this group at > http://groups.google.com/group/jbooktrader?hl=en. -- You received this message because you are subscribed to the Google Groups "JBookTrader" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/jbooktrader?hl=en.
