The Kelly Criterion is a money management and clutching system that is designed to maximise the growth of your bankroll over the long term.

Put another way, it gives bettors a method of calculating the optimal amount to bet on a horse and the best way to take advantage of overlays and underlays.

Developed by John Kelly working at AT&T's Bell Labs in 1956, it's been tossed around and mentioned in numerous handicapping books and always seems to pop up when the topic of money management comes up.

This article isn't going to delve into the mathematics behind the Kelly Criterion; rather, it'll try to discuss some of the strengths and weaknesses of this money management system as well as point out a few related web sites on the Internet.

HOW DOES THE KELLY CRITERION WORK?
The Kelly system works by leveraging the odds . . . finding spots where there are overlays and betting a portion of your bankroll based on how much of an overlay it is. By taking into account the expected rate of return and the risk involved, Kelly's utility function maximises the overall growth of your bankroll.

Betting more than the suggested Kelly amount is an undue risk; betting less than the Kelly amount will cause your bankroll to grow more slowly.

Of course, bankroll growth isn't a guaranteed thing. Using the Kelly Criterion successfully requires that you be able to find overlays when they arise and take advantage of them.

So to use the system properly, you need to be able to accurately assess each horse's chance to win a race and you need to be able to do it better than the crowd as a whole. If you can do this,  the Kelly system is the ultimate method for taking advantage of overlays.

So using the Kelly Criterion properly means that you'll need to create your own morning line ahead of time to get an idea of how you feel the race is going to shape up.

For instance, let's look at an imaginary three-horse race; let's assume you've done the handicapping ahead of time and come to the conclusion that horse No. 1 has a 30 per cent chance to win the race, horse No. 2 has a 50 per cent chance to win the race and horse No. 3 has a 20 per cent chance to win the race.

However, on the tote board, the crowd has given horse No. 1 a 25 per cent chance to win the race. Using the Kelly Criterion, you'll be able to figure out what percentage of your bankroll you should wager on the No. 1 horse.

While that may be a simplification of the system, it's essentially how it works: bet when you spot overlays, hold back when you don't.

With the Kelly Criterion, you are always betting percentages of your bankroll - as your bankroll grows, so do your bets. Likewise, when your bankroll shrinks, your bets will shrink.

The suggested percentages to wager can vary drastically depending on the expected rate of return on a specific bet, but you'll never find that you'll need to wager your entire bankroll on a single horse, nor a large portion of it unless you spot a huge overlay.

The maths behind the system is pretty complicated. Kelly's original paper is all but unreadable to those not big on maths. Some useful Internet website links for figuring out the mathematics of the system and a calculator which will do all the maths for you are provided on page 39.

WHAT ARE THE PROBLEMS WITH THE KELLY CRITERION?
There are a number of problems with the system, but none that are critical or make it impossible to use.

The Kelly Criterion works best when you know the final odds of all the horses involved. Since this isn't practical in tote betting, you'll have to make do with odds taken a few minutes prior to post time. It also assumes that you can judge each horse's chance of winning better than the crowd at large can (in other words, that you can accurately spot overlays and underlays).

To make money with the Kelly Criterion this is essential because the amount you are supposed to bet depends for the most part on the difference between a horse's actual odds and the fair odds as you perceive them.

The Kelly Criterion is also meant to be a long-term money management system. Rather than going to the track each weekend with \$100 in your pocket and betting it all, it's assumed that you are keeping track of your overall bankroll for a long period of time, not replenishing it with new money each time you head out to the track.

For most handicappers this is a complete change in the way you deal with your money and it'll require you to set aside a bankroll which you'll need to stick with for a while. Using the Kelly Criterion any other way is pointless because it defeats the purpose of Kelly's system, which is long-term bankroll growth.

The Kelly Criterion also becomes less practical to use when your bankroll gets smaller.

In his original paper, John Kelly assumed that you could always bet a percentage of your bankroll, even when that percentage amounted to less than a dollar. Mathematically this works well, bet 50 cents of the one dollar remaining in your bankroll, if you lose, bet 25 cents, etc.

However, at a track that only allows \$1 minimum bets you'll quickly realise how silly this is. So if your bankroll goes down to a few dollars, it's probably a good time to make a mental note that your ability to spot overlays isn't as good as you thought it was and start again from scratch.

It's also very possible that the Kelly Criterion will advise you to make a wager (albeit a small one) on a horse with a negative expected rate of return. In other words, a horse may be at odds of 40/1 on the board and you might perceive its fair odds to be 35/1, but you'll be told to bet a small amount on the horse anyway.

That's just the way the maths works and it's the exception, not the rule.

The Kelly Criterion also requires some preparation before heading to the track and a lot of maths. Many handicappers aren't used to creating their own morning line before going to the track, but if you want to use Kelly's method, it's a must.

Some simplifications of the system allow you to simply figure out how much to bet based on the odds of a horse and the odds you feel it should be at, but I wouldn't recommend using those variations unless you're too lazy to build your own morning line.

For most people, it's simply too much maths to do in one's head or to use a calculator at the track. As a result, many handicapping books have boiled Kelly's method down to simple statements such as:

Bet more money as your bankroll grows and less when it shrinks.

or ...

Bet a certain percentage of your bankroll

or ...

Bet more on a horse when you perceive it to be an overlay and less when it's an underlay.

While all true, these statements may do more harm than good because if put into use without underlying maths, they do no more than make you feel like you're using good judgement.

The actual Kelly Criterion allows you to quantify exactly how much you should bet on a given horse based upon its odds and what you consider to be fair odds. The above statements do no more than summarise some of the aspects of the Kelly Criterion.

While there are some mathematical simplifications of the Kelly Criterion that make it easier to use, they are also approximations and don't always suggest the proper amount to wager on each horse. To be used at the track, you'll need a computer program or programmable calculator to do all the number crunching for you.

SCALING BACK ON THE OPTIMAL KELLY WAGER
For some of you, maximising the growth of your bankroll may not be the most important objective. While you have to admit that it sounds appealing, isn't it just as important to ensure that you don't lose everything?

If safety is your prime concern, but you still like the idea of growing your bankroll, some studies have suggested fractional Kelly methods.

For instance, instead of betting the optimal Kelly amount, wager one-half or one-quarter of the suggested  mounts. Such strategies have the benefit of reducing the likelihood of you losing all your money. At the same time they'll tend to slow down the expected growth of your bankroll. It really depends on what's the more important longterm goal for you: growth or safety.

There are also times when the suggested Kelly bet may be larger than you can stomach. In these cases, it's sometimes suggested that you scale back just to be safe.

If a horse that you felt was a 2/1 shot to win the race is showing up on the tote board at 40/1, it might be suggested that you wager a large portion of your bankroll. Unless you are truly sure your 2/1 assessment is correct, it might be wise to wager a little less.

By Miles Michelson

PRACTICAL PUNTING - FEBRUARY 2002