Risk mgmt for short-term trading strategies?

bluelou · Post by **bluelou** » Mon Feb 02, 2009 4:31 am

Can anyone point me toward any academic research or books on risk management/portfolio diversification/position sizing for short-term/high frequency trading strategies?

I'm running strategies on futures with average holding periods of 6 to 36 hours.

FWIW, Here's what I'm biased toward using:

1) Simple linear constraints.

2) Vol-based and risk-of-ruin based criteria

Here's what I'm biased toward ruling out:

1) Using anything related to daily correlations doesn't seem to make much sense due to the short holding periods.

2) I'm weary of anything VaR-based b/c I don't trust the distributions - no matter what the methodology.

3) I don't have much interest in using Kelly-based criteria - I don't have that level of confidence in prior probabilities in financial time series. Regime switches happen far too often in this business.

Thanks in advance for your suggestions.

-BlueLou

Tradenator · Post by **Tradenator** » Mon Feb 02, 2009 5:09 am

See filthy's book on volatility, which has a set of measures you might find useful for your vol-based criteria. There is a thread here on the book so it should be easy to find.

jungle · Post by **jungle** » Tue Feb 03, 2009 12:18 pm

I think this topic was discussed recently; can't remember the thread, but FDAXHunter or Johnny might, as if memory serves they both contributed.

FDAXHunter · Post by **FDAXHunter** » Sat Feb 07, 2009 10:19 pm

Good memory Sir Jungle. I believe you are referring to this one:

[url=/Show%20Post.aspx?PostIDKey=115893]Risk management for high frequency trading[/url]

LowDD · Post by **LowDD** » Tue Feb 10, 2009 4:23 pm

So I read the prior thread and didn't find much thats useful there. Except to say that the better thinking is somewhat inline with the criteria you've already laid out. I have no problem with a well done VaR number (and your boss probably doesnt either).

You havent laid out enough information on: what futures? how many instruments? what is the frequency of actionable signals?

I think your "risk-of-ruin" is the right criteria. So this logically follows with something like 10% position size if you were just trading SPX futures (lame!) or crude futures (even lamer!) bc the crisis risk is something like 20-30% down. (I know you're going to say "but I stop out before that"... and then I say, "except when it matters"). The cool thing is the whole game is developing 10 more instruments/signals and then 10 more. So if you're playing the game well... the low position size quickly falls away.

Make sense? I know its not the academic link you asked for, but I really dont think academics have done a good job 'not blowing up'. So that might be a silly place to look.

A mantra: diversity is AS important as quality.