Reply

Book for education

5 replies

Fluke

Subscriber, bbp_participant, community, 43 replies.

Visit profile

8 years ago #113948

Hi guys, in this days i have a question in my mind. What books should I buy for myself culture of genetics evolution? 
I want use strategyquant with scientifc metod, i want calculate, those data are used to a certain type of strategy? And things like that, but calculated analytically, and not empirical.
Someone, or Mark himself, knows of good books to read on the subject?

I would appreciate even a small response, it is the third thread I open without even the slightest sign of life.

0

mikeyc

Customer, bbp_participant, community, 877 replies.

Visit profile

8 years ago #131325

Personally I don’t use the genetic evolution part of SQ much.  I find the random approach finds the best strategies, then improve parts again random. 

 

To me the genetic building just finds many clones that are all very similar and highly correlated.  I use it, but only for a short time to avoid clone making….

 

Just my 2 cents on the approach.

0

Fluke

Subscriber, bbp_participant, community, 43 replies.

Visit profile

8 years ago #131327

ok, but even if you use the random approach, how you calculate the data range? haphazard? empirical?

0

mikeyc

Customer, bbp_participant, community, 877 replies.

Visit profile

8 years ago #131328

ok, but even if you use the random approach, how you calculate the data range? haphazard? empirical?

 

I’m not sure I understand what you mean by data range?  Period of in sample to out of sample data?

0

Fluke

Subscriber, bbp_participant, community, 43 replies.

Visit profile

8 years ago #131335

the total period of data, is+oos

0

mikeyc

Customer, bbp_participant, community, 877 replies.

Visit profile

8 years ago #131337

Well, if you choose a very short period of data, for example 1 yr, you get better results, since the strategy is only “seeing” a limited range of market conditions and will adapted to these recent .  If you use a very long period of data, say 10 years, you cover most market conditions, but you get less relevant results (the market was quite different 10+ years ago) but a more robust strategy since it would show profitability over different conditions.

 

What is a scientific way to choose this?  I don’t know. It’s a compromise between creating strategies that are more intune with recent market conditions vs ones that are more flexible but are likely on average to be less profitable.

0

Viewing 5 replies - 1 through 5 (of 5 total)