Monday, September 1, 2008

Optimal f - Betting the Farm


Optimal f is the optimal % risk that can be applied to a fixed fractional money management scheme that will yield the greatest net profit. Of course for the net profit of optimal f to be positive the expectancy of the strategy must be positive. This method is also known as the Kelly criterion.

The primary disadvantage of optimizing our risk parameters in this fashion is the high degree of volatility that this will incur in your account. As such, this method utilizes no limiting factors to account for things such as margin calls. Also, it fails to keep risk within human psychological boundaries as when a trader experiences an 95% draw down (which could occur in 5% of all trade sequences) it is unlikely they will be able to keep trading in this fashion. Lastly, and if the above reasons were not sufficient, this methodology assumes a constant statistically verifiable expectancy. In other words - in the real world where trading expectancy is not constant, optimal f is not constant either. Which in layman's terms means that you are constantly trying to catch a falling dagger - drunk, blindfolded, and missing 3 fingers. Hence, it is an interesting theory but of little practical relevance.

LT

8 comments:

Anonymous said...

We should have a great day today.

Anonymous said...

LT, I couldn't agree more with that last sentence.

TLT

Lord Tedders said...

TLT,

I think the key to utilizing these types of analytics is to always think in worst case scenarios.

One use for optimal f that actually is useful is comparing optimal f to the standard 1-2% fixed fraction. If optimal f is say 6% I know that my system has very little room for error. On the other hand if it is at 30% I know I have a more robust method (assuming my sample size is statistically significant.

Anonymous said...

Well, that's one way to look at it! ;)

Anonymous said...

hi,how did u suffer a 95% drawdown if you were only betting 5% at a time?

Lord Tedders said...

Read the equity curve chart again lol. The Kelly Criteria is 34%. Hence the 95% drawdown. And this is just a SIMULATION I would never never risk that much on a single trade.

LT

Palm said...

I know the level of drawdown is insane. But the possibility is also very promissing. I'm trying to test out the idea. I'll allocate some tiny amount to a system and use optimal f to trade it. I will keep betting with the F and stick withit even if this account grow to be significant portion of my total equity. If the system stop working, then i'll loose the allocated portion. But if it doesn't, meaning that the system still has an edge over next 4-5 years, i might witness the greatest returns-per-dollar-invested investments. Something i can talk about at a party. I'll come back to update, hopefully in 4-5 years.

Ralph Vince said...

Gentlemen,
Optimal f is not the Kelly Criterion -- the two give very different answers, and the latter should not be used in trading. See my paper on this at www.ralphvince.com on "Optimal f and the Kelly Criterion."

Second point -- it IS entirely possible to have a very good idea where the optimal f point in a future time window WILL be, and to minimize the effect of missing what is the optimal in the future.

Yes, it's a rough ride regardless. -Ralph Vince