The key part there is 'educated guess', the optimal F doesnt leave you much room to be incorrect(by definition) and you are broke. Plus there is all the fat tails and things like that
Hello Mr. Vince, would you be kind enough to respond to comments below that your work is unrealistic since optimal f depends on largest loss, which is in general unknown? http://www.turtletrader.com/optimal-f.html Thank you in advance. Tim
I think you are too generous. It can get you broke in the short run like %Kelly. IMO the key to profitable trading is not how to allocate optimally to one instrument but how to use the lowest possible allocation to trade a number of uncorrelated instruments. It is math. If you are trading a single instrument and you are optimally allocating to get an expectancy Eo at high risk, you can get the same expectancy Eo = NEi, where N is the number of instruments and Ei is the expectancy for each with low risk, something like 0.618% of bankroll, a figure that I like.
The largest loss is only important so as to bound your answer (for optimal f, a fraction between 0 and 1, unlike the kelly Criterion answer, which is a leverage factor between 0 and infinity). By bounding your answer between 0 and 1, you open up all kinds of very favorable things. However, you will get the same number of contracts to trade, invariant of your biggest loss, with Optimal f. So, as I say, the largest loss simply bounds the value between 0 and 1. The arguments people volley against the notion of Optimal f -- e.g. too sensitive to biggest loss, cannot know what value to use in the future, etc., are minor and easily overcome. I'm not so much trying to sell people on the idea anymore, as I am enlighten them about what's going on with their trading and how these principles are germane to what they are trying to do. But it isn;t something I profit by (having others accept the idea that is) -Ralph Vince
goodgoing, Kelly is not a %. Using it as such will REALLY get someone into trouble, It's not the same thing as the Optimal f. Let's suppose I have one market where the Optimal f is .25. Do you think you are at any less risk trading 25 markets at .01 where the correlations between all pairs of the 25 components is 0? Don't delude yourself. This is a CLASSIC case of the disconnect between classroom and actual risk. Diversification does NOT mitigate risk -- reducing net positions size does. -Ralph Vince
I can't imagine a situation where a valid Kelly formula (yes, there are invalid ones and you should be aware of that) would have you allocating 80% of your capital unless you have a near-Grail trading system. Be that as it may, Step 1: make sure your Kelly calcs are valid by learning the real Kelly formulas from this tutorial: http://www.financialwebring.org/gummy-stuff/kelly-ratio.htm Or, if you want to cut to the chase, check out my thread. Step 2: if your tested trading system already includes trading multiple instruments (say n instruments), DO NOT TRADE n times the kelly fraction!! If you're basically using the same strategy on n separate instruments, you must insure that they are completely uncorrelated before even considering adding the separate Kelly fractions (and the chances of five independent instruments each yielding a Kelly fraction of 80% is somewhere between null and void, unless you do have a Grail or very nearly Grail trading system). As pointed out in the above links, the simplistic Kelly formulas will readily lead to overbetting, which is where I'd bet your 80% figure is coming from. Learn the true Kelly and avoid the pitfalls of oversimplification in your Kelly calcs. Oh yeah, avoid optimal f; it ain't telling you a thing that isn't already covered by true Kelly.
I like to look at this thing.. http://www.hquotes.com/tradehard/simulator.html One fun thing I can do is to use the default settings and simulate 100 accounts and imagine the guy with the best account saying "I have the grail!!" and the guy with the worst account saying "trading is rigged, the little guy can't win!!"
So what? This doesn't mean that the guy with the best account made money by chance. He may have an edge. The simulation just says that given a large number of market players who bet optimally, their equity curves will be normally distributed according to their win rate and R:R. That is all. Nothing spectacular. You don't need a simulation to know that.
kut2k2, Kelly and Optimal f are different indeed, and Kelly does NOT yield the optimal fraction. If you think .80 as a result of the Kelly Criterion means you you have .80 of your stake at risk, you are completely wrong and don't understand what the formulations are returning to you. To Understand this, see the recent IFTA Journal available at admin at ifta dot org. You're giving advice to others about this here based on incomplete and inaccurate knowledge of the matter. -Ralph Vince