|
Track Tracts
Variable Percentage Betting:
Better than Kelly?
by Gordon Pine
In the hope of squeezing another percentage or two of profit out of
their handicapping, horseplayers have used all kinds of betting systems. Betting systems involve changing bet sizes depending on certain rules.
There are all kinds of schemes out there, including things like due column wagering, progression betting, base bet plus square root and
more. Most of these are fruitless - some are dangerous to your bankroll.
These methods are so useless that it is a common precept among
money-management experts that all betting systems other than the Kelly method are futile. For the uninitiated, the Kelly method advocates
betting a percentage of your bankroll. The percentage is calculated by taking your edge and dividing it by the applicable odds. Some bettors
calculate their edge on every bet by comparing their betting line to the track odds. Others look at their historical betting performance and
divide their long-term edge by their average odds.
Barry Meadow, in his excellent book Money Secrets at the Racetrack
says, "None [of the betting systems] are as effective as the Kelly method." Other money management experts reiterate that point. Some
go as far as to flatly state that 's mathematically impossible to improve upon your flat-bet expectation using betting systems.
Is this true? Is it futile to try to increase your profits by changing your
bet size, except with the Kelly method? A guy named Jerry Samovitz disagrees. In his booklet Out of the Red, Into the Black, he promotes a
betting system which he claims can improve your Return On Investment (ROI). Without giving away the entire plot, his method involves betting
a percentage of bankroll. However, the percentage is not fixed. It varies within a range and is raised or lowered depending on recent wins
or losses. The idea is to grab the money when you're in a winning streak, and cut your losses as much as possible while you're losing.
Okay- the experts will tell you that this won't help. But being of an
iconoclastic bent, I thought I'd test Samovitz method anyway. And guess what? In my case, he was right. I created a program that went
through a database of thousands of "imaginary" bets I had recently made while testing several methods. Using Samovitz recommendations,
the ROI improved by 9%. It went from .85 with flat betting ((that's -15 cents per dollar wagered) to .94 with variable percentage betting (-6
cents per dollar wagered). (Thank heavens they were imaginary bets, not real ones.) By fiddling with the parameters, I could increase the ROI a few more percentage points, too.
Okay - the experts will tell you that one sample doesn't prove anything.
And they're right. But it is empirical data, and it definitely isn't a vote for the view that all betting systems are futile. At the least, it shows
that variable percentage betting deserves a closer look before being dismissed.
Hopefully there is now a crack in your mental bulwark that previously
had kept out all the bad bogus betting systems and prevented them from polluting the pristine purity of Kelly wagering. Now that you've
seen demonstrated, in fact, why the conventional knowledge regarding horse race betting systems may be wrong, I'll tell you why, in theory, it
IS wrong. All the fancy math that goes into proving that it's fruitless to adjust your betting amounts is based on an assumption. This
assumption is that each event (bet) is independent. Now, if you were tossing dice or playing roulette, this would be true. It wouldn't matter a
whit where the ball landed last time. Each spin of the wheel is independent. Your chances are the same each and every time, and any
betting system you used (besides Kelly) would be useless at best.
"All the fancy math that goes into proving that it's fruitless to
adjust your betting amounts is based on an assumption."
But this is horse race handicapping. Each bet that you make is NOT
independent of the ones that preceded it. And why is that? Because you are human, because you make judgments about each bet, because
what you're doing correlates with the reality of the track at times (usually coinciding with a winning streak) and doesn't at others (usually
coinciding with a losing streak). Because some days you feel good and make clearheaded decisions and other days you're a dope.
This is not to say that randomness doesn't affect a handicapper's
results. It does. But to say that the winning streaks and losing streaks of a horseplayer are just random variations, and that each bet he
makes is independent and has nothing to do with the handicapper's current sharpness and conformation to the current realities of the track
is just wrong. And it is this false assumption that brings down the house of cards that says, for horseplayers, Kelly wagering can't be improved
upon. It can, because horseplayers aren't dice.
Having said all this, I need to regroup and state that I'm a strong
advocate of the Kelly method, as long as it's modified to fit the realities of horse race betting. I think variable percentage betting has
possibilities for improving ROIs, as long as it's used in a disciplined and systematic way. I'm definitely not advocating due column wagering. But
there is room for improved betting methods in games, like horse racing, that aren't dictated by pure chance. NC
Copyright ©2001 NetCapper Inc. All rights reserved.
|
