Question? Ralph Vince & How much to put on a trade?

Discussions about Money Management and Risk Control.
Post Reply
joeb4now
Contributor
Contributor
Posts: 2
Joined: Thu Feb 05, 2004 8:31 pm

Question? Ralph Vince & How much to put on a trade?

Post by joeb4now » Fri Feb 06, 2004 11:46 pm

I've been searching for an answer in the field of "What is the perfect amount to risk per trade - with all factors being accounted for"

I recently read Ralph Vince "The new money managementâ€

edward kim
Roundtable Knight
Roundtable Knight
Posts: 344
Joined: Sun Apr 20, 2003 2:42 pm
Location: Silicon Valley / San Jose, CA USA
Contact:

Post by edward kim » Sat Feb 07, 2004 12:11 am

Hi joeb4now,

Use the search feature on this forum and use Google to find answers to your questions about Optimal F, formulas, and books. When you are done doing that, you might be able to ask more specific questions later that were not previously addressed on this forum.

Edward

joeb4now
Contributor
Contributor
Posts: 2
Joined: Thu Feb 05, 2004 8:31 pm

Your Advice

Post by joeb4now » Sat Feb 07, 2004 1:50 pm

Thanks for your advice I will look harder, It's kind of funny you recommended Google. I found this forum while I was searching for the answer to the question I just posted.

When I found the forum I did search it for every combination of my question I could think of, and found things that were unrelated.


I think my question is pretty direct, I have been doing a lot of research on the subject, I felt it seemed this forum was a group of professional traders so I thought I may seek their opinion on the subject.

I've been to many seminars as well which gave out decent formulas on the subject but I was looking for something more.

I am currently using {Account balance x Risk Percentage}/largest Loss= contracts or shares to trade this is ok it's the one Larry Williams recommends but hopefully you guys have found something better.

leonardo
Roundtable Knight
Roundtable Knight
Posts: 100
Joined: Wed Apr 30, 2003 2:27 pm
Location: Minneapolis, MN
Contact:

Post by leonardo » Sat Feb 07, 2004 6:28 pm

joeb4now
I've been searching for an answer in the field of "What is the perfect amount to risk per trade - with all factors being accounted for"
I think what you are searching for is what everyone would like have too. The problem with the solution is the "all factors being accounted for" tag.

You will find the greatest help for answering this question by studying utility theory. (also very easily referenced by Google)

Utility theory explains why I will trade up to 5% of my equity on my trades, others 8%, and still others here grudgingly will trade .5% on the very same trades. And none of us are necessarily "wrong".

Best wishes.

Leonardo

stancramer
Roundtable Fellow
Roundtable Fellow
Posts: 60
Joined: Sat Feb 07, 2004 10:34 am
Location: Washington DC

Post by stancramer » Sun Feb 08, 2004 12:42 am

JoeB, it is always proper and often insightful to ask "What do YOUR test results show?" (abbreviated WDYTRS). Please tell us, when you try three or four or five different methods of choosing your betsize, what do your test results say? Is one "better" (i.e. you like it more) than all the others? Do they all produce similar outcomes? Or are they different, and you can't decide which you like the most?

How can we help you interpret your own results (abbreviated HCWHYIYOR) ?

If you have no test results, you are only going to get the standard boilerplate reply, Do Some Work On Your Own And Think About Your Test Results (abbreviated DSWOYOATAYTR).

May God Bless You.

Kiwi
Roundtable Knight
Roundtable Knight
Posts: 513
Joined: Wed Apr 16, 2003 1:18 am
Location: Nowhere near

Post by Kiwi » Sun Feb 08, 2004 4:24 pm

From a book reveiew:
trade (not drawdown) experienced in backtesting. If you use optimal f and the largest loss in actual trading is greater than the loss experienced in backtesting, you will go bankrupt. Vince deals with this problem in an offhanded manner by suggesting that the actual f you use should be "padded". OK, so in the end you don't even use the actual optimal f, you pad it. And how much do I pad it by? Vince is silent on this question. So the purpose of optimal f - to decide by formula how much capital to allocate to a trade - is totally negated by the fact that you must "pad" optimal f. And you must pad it by a qualitativly determined amount because, again, Vince gives no formula on how much to pad it by. Optimal f is totally useless for system traders or any other trader for that matter.
And here is a formula B(n) = (1+f)^W*(1-f)^L*B(0) which comes from a much more detailed discussion at http://www.bjmath.com/bjmath/Betsize/mlmax1.htm

Be careful out there (esp if u get close to optimal or believe ryan).

John

ksberg
Roundtable Knight
Roundtable Knight
Posts: 208
Joined: Fri Jan 23, 2004 1:39 am
Location: San Diego

Optimal-f

Post by ksberg » Sun Feb 08, 2004 7:18 pm

Joe,

Ralph provides BASIC code for finding Optimal-f in Appendix B, page 189. He describes two methods for find Optimal-f starting on page 84.

Optimal-f isn't succinctly described in a short formula. To quote one web source "Optimal-f is found in a round about way". The optimization involves walking a curve to find a top or maximum point, just before sliding down the other side of the curve. In terms of application, optimal-f is still fixed fractional bet size (BTW: If you look closely at Larry William's formula, it is also simply fixed fractional bet size).

There is a serious consideration to ask yourself: what do I mean by optimal? In Ralph's case, he spells it out clearly: he chooses optimal to mean the fixed fraction value that yeilds the highest geometric mean growth. In that regard he is 100%, spot-on correct.

In my opinion, "optimal" involves other aspects, such as draw-down and stability. So I might use something like Monte Carlo analysis with a constraint question of "what value of f would result in a confidence level where the probability of reaching a 50% draw down is less than X%". Selecting, say 95% confidence, I could derive the value of f and start testing with it over multiple time-frames, markets, etc. So, even MC analysis only provides a starting point. In the end, I've got to feel comfortable with the allocation risk.

Stan asks some good questions. It come down to doing your own homework. In my experience, there isn't a blanket answer for all sizing situations. The answer varies by system, techniques within the system, and certainly by the person. Cool thing about being human, I guess.

Kiwi, I don't know where you got the quote, but I'd offer that Optimal-f actually does have value to the trader. For one, no trader should ever be above the optimal-f line ... many times the Kelly formula will be (hint: maybe you don't want to use Kelly). Secondly, Optimal-f provides great insight into system performance potential and can even be used as a ranking metric between systems (along with MAR et al).

Cheers,

Kevin

Kiwi
Roundtable Knight
Roundtable Knight
Posts: 513
Joined: Wed Apr 16, 2003 1:18 am
Location: Nowhere near

Post by Kiwi » Sun Feb 08, 2004 9:02 pm

I liked the quote Kevin especially the logic that leads one to conclude that if u have to pad it then it is totally useless. :wink: My first hit searching on Google .

I felt it was rather nicely balanced by finally providing a formula for the long and patiently waiting joeb4now and, rather more useful, a link to a page that discusses the application of the formula (first hit with slightly differnet phrasing).

All proves Eck's point really.

I also like Leonardo's post and your points about taming optimality with montecarlo. My own additional warning would go along the lines of "the past is an imperfect predictor of the future."

John

lperepol
Senior Member
Senior Member
Posts: 30
Joined: Thu Feb 26, 2004 12:50 pm
Location: Castlegar, BC, Canada
Contact:

Post by lperepol » Fri Feb 27, 2004 7:22 am

leonardo wrote: joeb4now
I've been searching for an answer in the field of "What is the perfect amount to risk per trade - with all factors being accounted for"
I think what you are searching for is what everyone would like have too. The problem with the solution is the "all factors being accounted for" tag.

You will find the greatest help for answering this question by studying utility theory. (also very easily referenced by Google)

Utility theory explains why I will trade up to 5% of my equity on my trades, others 8%, and still others here grudgingly will trade .5% on the very same trades. And none of us are necessarily "wrong".

Best wishes.

Leonardo

For me Thorp explains optimal f the best in his paper, "Portfolio Choice and the Kelly Criterion."

The main problem is with three types of stats, descriptive, predictive and prescriptive.

Descriptive: Data on an observed behavior is fitted mathematically. Many different utility functions might be needed, corresponding to widely varying circumstances, cultures or behavior types.

Predictive: "Explains" observed data: fits for observed data are deduced from hypotheses, with the hope of future data will also be found to fit. Many different utility functions are needed, corresponding to the many different sets of hypotheses that may be put forward.

Prescriptive: A "guide" to behavior -- a recipe for optimally achieving a stated goal. Most traders use optimal f as a prescriptive stat -- a prescription for allocating resources so as to maximize the rate of growth of assets.

So to finding optimal f may require you to become a doctor. One who writes prescriptions.

Lawrence

Forum Mgmnt
Roundtable Knight
Roundtable Knight
Posts: 1842
Joined: Tue Apr 15, 2003 11:02 am
Contact:

Post by Forum Mgmnt » Fri Feb 27, 2004 7:55 am

There's also been much written on this same topic in:

viewtopic.php?t=177

Since that topic is specifically entitled "Optimal f" let's continue discussions of Optimal f there.

- Forum Mgmnt

P.S. I'm really not a fan of using optimal f. I think it is useful to know there is a relationship between bet size and profits and that there is a maximum but any numbers coming out of the formula are decidedly useless. See the above topic for my primary reason.

Post Reply