Negative Expectation Misconception

[jfdinneen]

In answering a question from a novice handicapper (private communication), it struck me how easy it is to misinterpret some fundamental mathematical concepts relating to sportstrading. In that light and for the benefit of other beginners on the forum, I wish to clarify the concept of negative expectation.

Some handicappers believe incorrectly that, if they play a negative expectation game, they will lose a percentage of their initial bankroll equivalent to that expectation. This is incorrect.

For example, assuming that the expectation of randomly betting favorites at Track X is -17% and that Joe Punter starts with a bankroll of $1,000, he expects to lose (randomly betting favorites at Track X) $170 (17% * $1,000) over a season. In fact, he can expect to lose 17% of his total bets (turnover), not of his initial bankroll. Suppose he makes 250 flat $30 bets on favorites giving him a total turnover of $7,500 ($30 * 250), he can expect to lose $1275 (17% * $7,500) - not $170! In other words, he stands to lose (on average) $275 more than his initial bankroll!

Apologies in advance if this concept has already been clarified by other forumites

John

[formula_2002]

I'LL MAKE A POINT OF SOMETHING.
IF YOU START WITH A VERY LARGE BANKROLL, IT COULD TAKE A VERY LONG TIME BEFORE A .83 ROI EATS IT ALL UP.

IN THIS CASE, A .83 ROI, WITH A $2000 STARTING BANKROLL ATE UP
10% OF THE BANKROLL

SO START WITH A LARGER BANKROLL, SAY $20,000 BANKROLL THEN, YOU WOULD HAVE LOST BUT 1% OF YOUR BANKROLL.

NOW THAT'S WHAT I CALL MONEY MANAGEMENT!!


from note #6 in

http://www.paceadvantage.com/forum/s...ad.php?t=45468

[singunner]

Few people realize the coin flips the other way as well. An ROI of 101% will be multiplied by the total amount of money you wagered, not your bankroll. For me, that's the most interesting part of this game. If they offer you 1% at a bank, you laugh at them. If they offer you 1% at the track, you laugh with them.

[Overlay]

Originally Posted by jfdinneen

Some handicappers believe incorrectly that, if they play a negative expectation game, they will lose a percentage of their initial bankroll equivalent to that expectation. This is incorrect.

Not to nitpick (since I agree with the main point of your post), but, despite the take, racing is not a true negative-expectation game. (That's what makes the concept of an overlay possible.) I think it's important for handicappers (especially new handicappers) to understand that, since it's one of the things that separates racing from most casino games and other forms of gambling where the player is fighting adverse odds that never change; any winnings are attributable entirely to luck; and losses are unavoidable for anyone who plays long enough. It's also why there's more to handicapping than just picking the winner.

[Cangamble]

If the average takeout rate was 40% it still wouldn't be a negative expectation game by definition, but at what point does one drop the conventional definition?

[Overlay]

As you say, the take is an obstacle that each player must overcome, and continually increasing it only makes consistent profit more difficult to achieve than it already is, to the long-term detriment of the sport itself. But even talking about the possibility of systematic, consistent profit has no meaning in the other games that I was referring to.

[highnote]

Originally Posted by Cangamble

If the average takeout rate was 40% it still wouldn't be a negative expectation game by definition, but at what point does one drop the conventional definition?

Good point. Dr. Z wrote a book "Dr.Z's 6/49 Lotto Guidebook" which shows how it may be possible to get an edge in lottery games. These games can have a 50% takeout!

[Norm]

A nickel's worth of input from an old-style 'capper. In the old days we were taught that horse racing was a winnable game because there were two sets of odds in play, the natural odds of an occurrence and the odds on the tote board. What you could win was the difference between the two, nothing more, nothing less. There is nothing else available to win but the difference. The take-out had no effect, what you saw on the tote board governed all decisions. And what, you should ask, were the 'natural' odds ? Well, that is what handicapping is all about and that is why handicappers are enthusiastic record keepers.

In real estate it's location, location, location. In handicapping it's records, records, records. It's a positive expectation game. Identifying the difference is how it's done.

[Cangamble]

Originally Posted by Norm

A nickel's worth of input from an old-style 'capper. In the old days we were taught that horse racing was a winnable game because there were two sets of odds in play, the natural odds of an occurrence and the odds on the tote board. What you could win was the difference between the two, nothing more, nothing less. There is nothing else available to win but the difference. The take-out had no effect, what you saw on the tote board governed all decisions. And what, you should ask, were the 'natural' odds ? Well, that is what handicapping is all about and that is why handicappers are enthusiastic record keepers.

In real estate it's location, location, location. In handicapping it's records, records, records. It's a positive expectation game. Identifying the difference is how it's done.

The problem is that spread between the natural odds and real odds is now at a point that it is pretty much unbeatable without a healthy rebate.
In the old days, and I know I've made this point over and over, there were no Beyer figures and there was a lot more blind public money (no slots and very little other gambling competition made it so the pools were full of dumb money, not anymore today). Now it is pretty much sophisticated handicappers against sophisticated bettor/handicappers.

[singunner]

Originally Posted by Cangamble

The problem is that spread between the natural odds and real odds is now at a point that it is pretty much unbeatable without a healthy rebate.
In the old days, and I know I've made this point over and over, there were no Beyer figures and there was a lot more blind public money (no slots and very little other gambling competition made it so the pools were full of dumb money, not anymore today). Now it is pretty much sophisticated handicappers against sophisticated bettor/handicappers.

I'm afraid I have to disagree. An informational edge can give you a huge advantage. Normally this takes the form of people actually attending a track and getting a feel for the horses or having inside information. Alternatively, you could find a way to do something that nobody else has done before.

With all the data collected about every single horse race, it's easy to feel overwhelmed. That's why computers have become so important. Beyer Speed Figures were great, but now that everyone has access to them, they've become worthless. What you need is something unique that hardly anyone else has access to. Modern handicappers have not even scratched the surface of what is possible with a sophisticated statistical analysis.

[Norm]

Originally Posted by Cangamble

The problem is that spread between the natural odds and real odds is now at a point that it is pretty much unbeatable without a healthy rebate.
In the old days, and I know I've made this point over and over, there were no Beyer figures and there was a lot more blind public money (no slots and very little other gambling competition made it so the pools were full of dumb money, not anymore today). Now it is pretty much sophisticated handicappers against sophisticated bettor/handicappers.

I have to agree that the difference between the sets of odds is a lot thinner than it used to be. In the old days profits were handsome, now-a-days profits are more modest even though the average serious player has increased his win percentage by many points due to the vast array of additional information. Today, I have to exercise a level of discipline that was unnecessary years ago, passing on many races that I would have played back then. I think the biggest obstacle for the modern player is not the 'take' (it would be nice if the 'take' was lower, but wouldn't that just draw more money into the pool on a low-priced horse ?) and it's not the sophistication of modern players (it's true, but no one here is afraid to lock-horns at that level) but rather the lack of technical sophistication in the quickness of reporting bets on the tote. Accepting a play at 6-5 because the reliability of the pattern supports it is still an overlay until, 20 seconds after the gate opens it becomes 3-5 and therefore a bad bet. I'm not suggesting post-betting, just that the systems are too slow. Oddsmaven has posted some research in another thread here that addresses this new-age problem. While there are still playable patterns available that get around this, the game would be far more profitable and offer a lot more plays if there was just a bit more certainty regarding final odds. I know you can never be 100% certain, that's not a reasonable expectation, but the bottom falls out after the bell all-too-many times these days. Wrestling with this one is the modern 'cappers biggest challenge.

[Cangamble]

Originally Posted by singunner

I'm afraid I have to disagree. An informational edge can give you a huge advantage. Normally this takes the form of people actually attending a track and getting a feel for the horses or having inside information. Alternatively, you could find a way to do something that nobody else has done before.

With all the data collected about every single horse race, it's easy to feel overwhelmed. That's why computers have become so important. Beyer Speed Figures were great, but now that everyone has access to them, they've become worthless. What you need is something unique that hardly anyone else has access to. Modern handicappers have not even scratched the surface of what is possible with a sophisticated statistical analysis.

The playing field has been too leveled. Sure, there are some angles that could be slightly profitable, but not enough when taking in margin of error over time.

The spread between the real odds and the natural odds now are too close.
Without a rebate, all I can say is good luck to you.
In other words you can't make money on flipping a coin over time if you also have to overcome a takeout.

[singunner]

I can understand why you feel the field is too level, but horse racing has become largely about technology. The thing with technology is that it always seems like it hit a wall, then someone surpasses everyone else's expectations and it becomes the new norm.

[GameTheory]

Originally Posted by singunner

I can understand why you feel the field is too level, but horse racing has become largely about technology. The thing with technology is that it always seems like it hit a wall, then someone surpasses everyone else's expectations and it becomes the new norm.

You mean like with SinStats? What do you mean by "competitive connections" anyway?

[singunner]

I guess since I'm not selling anything yet, maybe the moderators won't get mad at me. Rereading my posts, it sounds like I'm pushing what I've made, but I don't think what I made is the only way to success. In fact, I'm sure there are a lot of ways to get a positive ROI that haven't been discovered yet.

Regarding SinStats, it's the project I've been working on for the last year. It uses an advanced statistical analysis algorithm to look at how horses compare based on a single variable. That is to say, it doesn't look at trainers or jockeys or running patterns or pace or whatever people are using to handicap this week. I don't make any crazy claims about it. All my output for the last two years is downloadable in Excel files on the site.

I really didn't want to get into this without being an advertiser, but my views are highly influenced by my own experience working on this project. I am one of those "whiz kids" you used to hear about, and horse racing has become a very interesting area for me to focus on. It's hard to separate my opinions from my research.

This post feels way off topic, and I apologize if I am out of line.