Sean-
My post references the book, "The Mathematics of Money Management" by Ralph Vince and the comments to your post reference this text.
One of the biggest complaints about Optimal f is the one you reference in item 1) a. below. Vince addresses that in chapter 4, "Parametric Techniques on Other Distributions", where he shows how to create a distribution of past trades and how to obtain the Optimal f of that distribution. Instead of using the largest loss you would end up choosing appropriate standard deviations off the mean of the distribution. At the end of the day this gives a probabilistic technique and avoids the "largest loss issue".
He also addresses the issue of the largest loss in Chapter 5, "Multiple Simultaneous Positions", where he shows a technique using options where you always know how much your largest loss is.
Items 2) a. and b., below, are addressed in Chapter 7, "The Geometry of Portfolios", where he shows how to create an optimal portfolio of tradeable instruments.
As for Item 3), below, Vince brings to the table that you cannot use a Bernoulli technique, as in the Kelly formulas, on a non-Bernoulli distribution, as are the returns in the stock market( see page 29 of the text that I referenced). So while he didn't invent the idea of geometric growth, I'd assert that neither did Kelly; that honor probably started with nature.
I don't encourage or discourage the use of Optimal f but I believe that some money management system should be employed.
I welcome any and all comments to my post.
...howard feldman
I can perhaps add to this discussion.
1.) Optimal f is a fraction of largest (hypothetical) loss in your backtest.
a. The problem is that the future results may and probably will have a greater largest loss.
2.) Assumes one traded instrument with the ability to adjust bet sizes between trades.
a. This process does not take into account concurrent trades (bets) across more than one instrument at a time, ie., portfolios
b. There is an efficiency of say adjusting bet sizes between 2% max risk per trade on one instrument (by itself) AS COMPARED to the anti-martingale efficiency of say 10 simultaneous trades of 2%, where if they all got stopped out at the same time (loss of -20%) before any trade sizing risk adjustment.
3.) There has always been some different plays on the Kelly formulation by individuals who alter it, rename it and call it their invention. But in reality they didn't invent anything.
4.) So, multiple concurrent bets involves "portfolio heat" and while you could employ some type of Kelly or optimal f on that, you still have the same issue of the future matching history and how close to the edge of ruin you wish to play.
-S
From: amibroker@yahoogroups.com [mailto:amibroker@yahoogroups.com]
Sent: Monday, January 25, 2016 12:44 PM
To: amibroker@yahoogroups.com
Subject: Re: [amibroker] Optimal F
This response does not answer the original post but I think I have something to contribute to the subject of Optimal f and may be of interest to the group.
I have a great deal of respect for the work that both Howard Bandy and Ralph Vince have undertaken.
The Internet is filled with detractors of Optimal f. I believe that most people have not labored past the first few chapters, those chapters that describe an empirical method for finding Optimal f. Those who have read the entire book find a few more techniques that are more in line with reality.
That being said, I agree with Howard Bandy's response to the original post. And I believe that Ralph Vince would agree as well. I recently corresponded with Mr. Vince about a system I was exploring that had an Optimal f of 0.80 and a worst case trade of 4% loss. When I did all the calculations I found that I would have needed more than 13 times my existing equity to trade at Optimal f, a highly impracticable amount. Ralph wrote back to me and explained that my figures were mathematically correct but that oftentimes we run smack into reality and cannot really trade at the level suggested by our system.
He also explained that if I was able to trade at my calculated Optimal f that when my system actually did encounter a 4% loss on a trade that the real amount of my loss would be a drawdown of Optimal f percent or in my case 80%. Candidly, I don't think there is enough Pepto-Bismol in the world to relieve that kind of pain. Such is the nature of leverage.
Ironically, trading a more modest system produces an Optimal f that is more practical to trade.
Does this mean that we should completely reject Optimal f? Absolutely not! We simply have to realize that Optimal f gives us the optimal amount we should trade with a "risk be damned" attitude. Maybe not very practical, sometimes, but Optimal f does provide you with valuable information about the maximum amount you can expect out of your system.
Howard Bandy incorporates risk into the trading equation and dampens the volatility in a more practical sense.
Both works of research are important but it's important to understand what each is saying on the subject.
...howard feldmanOn 01/23/2016 04:38 AM, queiroz_everton@yahoo.com [amibroker] wrote:
Hello,
Does anyone have the AFL code of Optimal F by Ralph Vince?
Best Regards
Everton Queiroz
Posted by: Howard MITCHell Feldman <hfeldman@earthlink.net>
| Reply via web post | • | Reply to sender | • | Reply to group | • | Start a New Topic | • | Messages in this topic (9) |
This group is for the discussion between users only.
This is *NOT* technical support channel.
TO GET TECHNICAL SUPPORT send an e-mail directly to
SUPPORT {at} amibroker.com
TO SUBMIT SUGGESTIONS please use FEEDBACK CENTER at
http://www.amibroker.com/feedback/
(submissions sent via other channels won't be considered)
For NEW RELEASE ANNOUNCEMENTS and other news always check DEVLOG:
http://www.amibroker.com/devlog/
EmoticonEmoticon