| I'm sure some people know about Efficient Frontier, | | | | the portfolio volatility. |
| but I'm guessing that there are less investors that | | | | 2. To use Kelly Criterion, it requires knowing how |
| know about Kelly Criterion. So what is Kelly Criterion | | | | good you trade stocks (in terms of p & b). |
| and who is Kelly? Kelly worked at AT&T, and | | | | Obviously, if you don't know exactly how much your |
| published his original paper back in 1956. Its math is | | | | "edge" is, the Kelly betting amount will probably be |
| quite involved with communication and information | | | | off from the correct amount. Estimating and knowing |
| theory, mostly dealing with probabilities. However, | | | | your edge will be a much harder task than calculating |
| behind all the maths, there lies an astonishing result: | | | | the Kelly betting amount. |
| by placing bet amounts according to Kelly Criterion | | | | Despite the mathematical correctness of Kelly |
| (originally applied to horse-race gambling), one can | | | | Criterion, it is much harder to invest such in practice. |
| maximize the returns in the long term. Here is the | | | | Aren't there anything that we can walk away from |
| betting formula which has been tailored to stock | | | | such a terrific investing formula? Indeed, there is. |
| trading: | | | | Here is what I personally learned after investing |
| K% = ( (b+1) * p - 1) / b = ( b*p - (1-p) ) / b | | | | stocks for almost 10 years now. The riskier the |
| Win probability (p): The probability that any given | | | | stock/or entry point is, the less amount that you |
| trade you make will return a positive amount. | | | | should put in; the safer the stock/or entry point is, |
| Win/loss ratio (b) or odds: The total positive trade | | | | the more amount that you should put in. This is |
| amounts divided by the total negative trade amounts. | | | | exactly the spirit of Kelly Criterion that bet should be |
| If you think of b as the odds of b-to-1, payout of b | | | | proportional to your edge or your supposed |
| when betting 1 unit of money, the numerator is | | | | advantage. I have been burned by stupid bets so |
| simply the mean value of expected payout, or the | | | | many times that I finally learned to carefully size each |
| so-called "edge". Therefore, K% can be expressed as | | | | of my stock transaction. In fact, sizing of your |
| edge/odd. For obvious reason, you don't want to bet | | | | transaction is equally important if not more than what |
| in any game where the expected payout is 0 or | | | | stocks you pick. While most of the investment world |
| negative. | | | | talks about what to buy, much less attention is spent |
| If Kelly Criterion is so great, why is that this is not | | | | on how much one should buy. But for every |
| heard or used very often in the investing world. | | | | transaction, it always consists of the following |
| There are a couple of reasons that prevent it to be | | | | elements: what (stock) to buy/sell, when to buy/sell, |
| used practically: | | | | and how much to buy/sell. For successful investing, all |
| | | | three elements must be carefully chosen. And Kelly |
| 1. The volatility of strictly using Kelly Criterion is quite | | | | Criterion helps you on deciding the last element: how |
| big. Despite that in the long term, probabilistically | | | | much. |
| speaking your portfolio will have the maximum return | | | | For more related articles, one can check out the |
| possible, the ups and downs are too big to be | | | | article from investopedia. Tom Weideman also has an |
| digested by most people. Therefore, people talk | | | | excellent article using simple calculus for deriving Kelly |
| about using "half Kelly" or half of the bet amount | | | | Criterion with less math from information theory. You |
| calculated from Kelly Criterion in attempt to reduce | | | | can find the original Kelly's paper here. |