Reply

To Optimize or not ?

3 replies

mx55333

Customer, bbp_participant, community, 84 replies.

Visit profile

7 years ago #112207

I notice that Monte Carlo is faster than WF Matrix. I know that WF Matrix will tell me if and how I should optimize strategy.

 

1. Here is how I am using Monte Carlo to decide if I should optimize:

 

If I run 50 Monte Carlo runs for the strategy (only checking ‘Randomize Strategy Parameters’) and see the AR of all the runs vary greatly, then I assume that the strategy is a good candidate for Optimization.  Is this a good idea to assume so OR should I stick to WF Matrix to decide if the strategy benefits from optimization.

 

2.  Should I even optimize the strategy, given that It performs well on preceding and succeeding OOS periods?  What are the dangers of NOT optimizing?

0

kanon103

Customer, bbp_participant, community, 46 replies.

Visit profile

7 years ago #124698

everyone is gonna say yes, do it is better than not to do it.

0

BladeRunner

Subscriber, bbp_participant, community, 24 replies.

Visit profile

7 years ago #124702

1. unless you use fix spread, you should monte carlo test also on randomizing spread and also slippage. If the montecarlo tests are also good with those parameters, it should be ok too into real trading.

 

2. optimizing already good strategy got pros and cons, so it only depends on your feelings about what should be or not optimized.

0

matka

Customer, bbp_participant, community, 189 replies.

Visit profile

7 years ago #124703

I think Randomize Strategy Parameters MC test can be interpreted only as indication how results vary with different values. I would use original values in live trading. I can’t see any dangers of not optimizing http://en.wikipedia.org/wiki/Overfitting

0

Viewing 4 posts - 1 through 4 (of 4 total)