Update: There is a more defined way to use the kelly criterion in investing that I will get to later, but it involves bet sizing with multiple potential outcomes that aren’t limited to “lose” or “win”. Check out the link on bet sizing and make adjustments. In reality losses will be generally limited to 8% rather then losing everything, and losses can be more severe overnight occasionally, and wins can be greater than 20% as well. Just because this update is here doesn’t mean using the kelly criterion is outdated.
There’s a model testing using the kelly criterion that shows the results will work just fine.
Think about it for a second. If the odds are not in your favor, it’s not worth betting. If the odds are in your favor it is worth betting. However, even if the odds are in your favor, if you take leveraged risk and the risk is too great, regardless of the fact you expect to win more than 50% of the time and return more than 2:1 on your money, you can’t win if once out of ever 100 bets (or trades) you lose everything. Some people believe if they only make that bet once and then they are more conservative after that that it is somehow a prudent decsion that will provide good results. That’s not true, and those are the same people that try to chase losses downwards and find themselves making risky bets often enough in their life that eventually they do lose everything. They convince themselves that it’s the “last time” many more times then they intend on it.
However, if you bet too little, aside from the fees and commissions eating into your profits, you won’t return enough to be able to have the kind of gains you would otherwise have.
It is essential that when you trade, you do so with the odds in your favor or else you are throwing money away.
So how do you determine this?
First of all you have to understand how to estimate how much you win when you are right, how much you lose when you are wrong, and how likely or how often you are going to win.
A lot of people like to believe their potential upside is infinite, and there downside won’t exist as long as they hold the stock long enough and eventually it will go higher than they bought it. This is rarely true, and there is ALWAYS a chance that you lose.
You wouldn’t sit down at a poker table to make a living without knowing everything you possibly can about the odds in every situation or at least how to figure them out, and how to make an informed correct decision in every situation to provide you with a mathematical advantage, and/or to limit your opponent from having an advantage on you. So why would you invest without knowing your advantage?
It is my sincere belief that most novice and newer traders are actually gambling more than professional blackjack counters and poker players. At least the poker players know their odds.
In chess, all moves are known. The combinations of moves and possible moves are all there to see. The only thing that’s not known is exactly what your opponent is going to do. In poker all possible outcomes still exist, but are not known, however probabilities are fixed which allows you to manage your decisions to produce long term returns by estimating the probabilities of events occuring. You can still estimate probabilities accurately by determining the probability that your hand improves, and the probability that you win if your hand improves. In stocks, it is more similar to poker than to chess. However, there are not a fixed number of stocks that get drawn from a lottery to determine how much each goes up or down. It’s more behavioral like both chess and poker, but it requires much more caution.
Like poker, you can estimate the probability of stocks going up based on past behavior, but unless you restrict yourself to buy and sell rules, it is impossible to determine how much you win and how much you lose in what amount of time. In poker, the rules are already defined and the number of cards are set. The amount you bet, you and your opponents determine, but the ratio you win to the amount you lose when you lose is pretty well set. If you bet $100 and you win, $100 is added to what you win. If you lose, that $100 is lost. It’s not as if you can put $100 in and just wait long enough for the pot to go up on its own without you putting anythin else in, aside from the amount that your opponents match you.
Stocks are different. In poker, you are fixed to a specific hand and you either win lose or split the current pot. In order to manage money properly when trading stocks, you must have a clearly defined “loss” (at what point you will sell if stocks go downward.) A clearly defined “win” (at what point you will sell if you are up.) Only then can you monitor past formations and moves to estimate your probability of producing a “win”. Only with the information of your ratio of how much you gain in a win to how much you lose in a loss along with the probability of winning and losing can you properly determine at what point you cannot risk any more without long term destruction of capital due to large swings that are impossible to recover from. Only at this point can you clearly identify how much you bet.
Keep in mind that history is not neccesarily an indicator of future, so even the probability is not truly known. As such, you should take additional steps to make sure you bet onthe more conservative side.
Additionally, a stop loss will not protect you against an overnight drop, and that is reason to be more cautious about your position size.
When you are trading, you have to have a very well educated guess at the odds of a particular trade. I suggest Encyclopedia of stock pattern. This will tell you the probability of hitting your price target as well as the percentage of stocks that will retrace by 5% (they call the break-even failure rate). So to find the probability of a win you take the percentage meeting the target minus break even failure rate.
This particular method of training is described in William O’Neil’s how to make money in stocks.
You define your loss as 8% by setting a stop. You define your win by looking at the chart pattern and determining the average gain as well as looking at the price target. I would always aim for a higher price target than the average gain produces, and for purposes of safety, only use the lower one when determining how much to put at risk. However, it is still okay if you use the price target and ignore the average gain made by stocks showing that pattern, it’s just more aggressive and slightly less prudent. dditionally, I prefer to find stock patterns that are within 5% of support and I will sell the stock only slightly below support resulting in closer to a 5 or 6 percent loss or less to add additional safety. I also assume worst conditions, so if I am buying a stock and it is going to hit it’s target 60% of the time in a bull market but only 48% in bear market, I asume as soon as I buy we go into a bear market.
Now how much you bet depends on what the money management tool tells you. In this case a “kelly criterion calculator“. And because the result is unknown and betting too much could be devestating, I would risk 1 half the kelly.
Technically this is how much you can risk to lose on a single trade. (not the value of the actual trade. So you would probably make sure your stoploss doesn’t lose this much) However, it assumes bets are made dependently. If you want to make multiple trades your total risk has to add up to this amount. Additionally, the real amount you can lose is much more the stop however the probability of a loss larger than your stop is not as good. If you really want to be wsafe, you either want to measure your results if you were to buy call options assuming you lose the entire option when you lose. Otherwise you will have to buy a protective put with a strike price at the stop price, figure that into the probability of profit and ratio of profit as buying puts will cost you in the size of profit. When the stock goes beyond the put price, you don’t need a stoploss, and instead should sell both the put and the stock itself. If the result is that you determine the kelly is 20% of your total capital, and you want to make 5 trades, each trade must have similar numbers and you must only risk 4% or less per trade. This way the total amount risked is 20% assuming that you have puts at the part where you would risk 4% of your capital, or that you have 80% cash. If you plan to make multiple trades with stop losses, you can use a little bit more because the chances of every single stock losing everything is very rare.
Example trade: you see a symetrical triangle forming with a price target 35% higher than the breakout point. You look it up and you see the following stats
RESULTS SNAPSHOT
Upward Breakouts
Appearance: Price forms lower highs and higher lows following two sloping trend lines that
eventually intersect. The breakout is upward.
Reversal or continuation Short-term bullish continuation
Bear Market
Performance rank 7 out of 19
Break-even failure rate 7%
Average rise 26%
Change after trend ends –33%
Volume trend Downward
Throwbacks 55%
Percentage meeting price target 57%
Surprising findings Busted patterns perform better than the originals. The best performance comes from
patterns near the yearly low. Throwbacks
hurt performance. Tall and narrow patterns
perform better than other combinations.
Plug the numbers in
http://www.albionresearch.com/kelly/default.php
Payout Odds are 26(upside average) to 8 or 3.25 to 1.
Probability of winning is 50% (57% minus break even failures of 7%)
•According to the Kelly criterion your optimal bet is about 34.62% of your capital.
If you chose to make this single trade, you might realize that a put will cost you 3% and commisions overall will cost you 1% So you would have have to go back and remove this from your upside and put in 22 to 8 payout odds. The Kelly criterion would then be 31.82%. You would take less than half of that or about 15% and that would be the total amount you should risk including put and commission cost. You would hold the put for 2 months or 8 weeks. You may sell the put before it expires allowing you to have a higher payout odds, but its’s always best to error on the side of caution. Since your stop is 8% plus the 4% and that’s 12%, you could actually put more than 100% of your capital at risk if you used protective puts. If you did not, you would only be able to put the 15% of the capital on a single trade. If you used multiple trades with the same odds without puts, you may decide to use 20% of your capital.
If you were to use call options, you would have to figure out the upside minus the time value of the option.
If the stock is $100 and the option costs $5 for at the money option that $5 is your cost. If you expect to have the stock to go to $126 before expiration your option will go to $26. Your payout odds are 26 to 5 or 5.2 to 1 However when it gives you the kelly this is actually the amount you risk since a loss is 100% of your bet not how much you lose if you “stop out”. The kelly will actually be how much you risk in a single trade,and you also will have to have the rest in cash.
Some may argue that you can also make bets against different stocks by buying puts because those bets are dependent. However there is a correlation and that means if the markets trade sideways, both options are more likely to expire and therefore the bets are not independent which is required for the kelly to be a useful tool.
You should never in my opinion risk more than one half the kelly on a single trade. You can probably make exceptions if you trade 100% of the stock without puts and instead owning a stoploss as it is very rare that you will stop out, but personally I do believe that the money is still to some degree at risk. Personally I think you could probably treat each bet as an independent bet if you use the stops but the truth is if you want to use this tool accurately you have to assume that all your money is at risk. It’s not 100% useful when your loss is only a fraction of your win. This is more useful if you use options and assume you are betting it all in an all or nothing situation. In reality there are new options that allow you to make all or nothing bets, and these would be perfect for the kelly criterion if you knew the exact percentage. The truth is it’s a game of making educated guesses and erroring on the safe side. I would say buying call options while owning additional time value on them and selling them short of expiration date and selling them if the stock goes below the stop would be a way to have your option maintain a lot of it value which will improve your true payout odds, while still trading within the confines of understanding that you could lose it all, would be the most accurate route but you do have to undestand that the past is not always an indicator of the future, and I would advize probably making much smaller bets individually, but none of this should be taken as actual financial advice as the terms of this website clearly explains. However it may be useful information that those who already have a sound system may decide to use to inform themselves to make their own investment decision, and I do believe that it is useful information for many.
I believe that money management is the most important decision. I can conclusively say that if you violate the kelly criterion by betting significantly more than the kelly criterion indicates and risk more that the chances are that unless you are lucjky and the numbers are more favorable than you thought, you will eventually lose everything. So this information should at least be able to protect those who are reckless from taking on too much risk. If you are taking less risk than the kelly indicates, you may not wish to increase the risk which is fine. It’s just a guideline one what you can choose to consider when making your own decisions.
Note:Investing in stocks in the long term (10 year period) is a different story, but there is significant risk. The risk is that the product won’t continue to be consumed, the company won’t grow, and the balace sheet will suffer, or that costs cut into the profit margins, or that corperate fraud makes the company go to zero. You can reduce a lot of these risks by ensuring you buy a irreplacable product with a “durable consumer advantage”. Even so, you still have to have your company with the earning power over the time you own it to be enough to increase it’s balance sheet, and continue to do so, or pay you a dividend so that you will either get your money back through the dividend, or that you will get the stock back by selling the companies stock shares at higher prices. However, this is going to limit risk, but not really define it unless you can esimtate the odds that the company doesn’t produce and that people don’t price it where you expect it.
The kelly criterion would still be applicable with the payout being whatever you believe the stock will be worth in 10 years to what you are currently paying. If you expect the stock to double in 10 years and it’s currently worth $60 the odds would be $120 to $60 or 2 to 1. If you believe there is a 50% chance of you hitting your price target, you may wish to revise that on the conservative end because people tend to be overoptomistic so a 40% chance. The problem is, the odds of a “win” may be 40% but if you do not win, you don’t neccesarily “lose” or have everything go to zero. So the Kelly criterion is not the perfect model in every situation but it can be used as a guide to make sure you aren’t taking on significant risk. You can also compare your return or average wealth increase that the kelly criterion estimates to the risk free rate of returns on US treasuries (with some risk factored in for risk of default or devalutation due to inflation.) Investing is a complicated game, but you arent simply buying produce for a week, you are investing a lot of money and putting your future at risk so make sure to not only do all the “homework” needed, but also make sure you look at all options and mae sure you can handle the risk.
edit: I just realize I was facturing the odds wrong. If the stock is worth $120 you will have gained $60 so they should be 1:1 in the example about long term investing. However it would remain correct if you define a “loss” as breaking even over 10 years.
Recent Comments