A Freakonomics Quiz for Horse Players

Here is a tough little Freakonomics quiz for people who like to bet the ponies:

Is there ever a situation in a parimutuel betting system in which you would want to bet on a horse to win, even though you knew for sure that the horse would lose the race?

This is a hard one. Had it simply been posed to me as a question, I don’t think I could have figured it out. I know of only one case in which someone actually followed this strategy, and it worked out great for them.

We’ll give some Freakonomics schwag to the first person who gets the right answer (unless that person’s name is Bill Hessert, because he is the one who first told me the story).


You are betting a sweep or pick six, and instead of correctly picking the winner of the last race, you'd rather let the jackpot rollover and win next time, thus collecting a larger payout.

Jeff Harbison

I suppose that if every single bet is on the favorite the payout would be about nill. However, a few bets on other horses would increase the odds for the heavy favorite from no return on a successful bet to a slight return. If you have enough action on the favorite the change in odds created by the losing bet might be economically justified.



One example is already given in the explanation article you linked to: arbitrage between the pool and an (outside) bookmaker. That said, arbitrage is less profitable than simple betting on your information if it is absolutely objectively certain that the horse will not win.


I'm not sure if this is what you're going for or not. If you place a bet early on, then it may affect others' later bets: People may not bet on a horse if no one else is betting on it, but when they see you bet on it, they may think there's something you know about it that they don't. Under the right circumstances, then, it could be a winning strategy to bet early on a loser, inducing others to follow you and increasing the total pool of money, and then bet on the winner just before betting closes.


If you bet on a true longshot in early betting, you could drive down the odds and hope other players follow the lead and also pick this favorite far more often than the underlying probabilities would indicate... which could result in the rest of the field collectively having a positive expected value.


It comes down to breakage (the house keeps the odd cents). If there is $199 bet total of which you are the only one who bet ($100) on the sure winner. You will get $1.90 per dollar or $190. If you bet one more dollar on the loser, you will get $2 per dollar bet or $200.

(You will not bet that extra dollar on the sure winner since the payoff will then be $200/101 = $1.90 when rounded for a total of $1.90 * 101 = $191.90.)


It would make sense if you were also certain that no money had been bet on the horse. Because their is an 'initial line' on the horse set by the track, putting any money on a horse that theoretically has an 'infinite payout' (or no payout as described in the attached article) would reveal alot about the pot odds, which would reveal a lot of information about how others are betting on other horses at that (fleeting) moment in time.


OK, the horse is old and crappy. The jockey a loser. Nobody, i said nobody, had put a single buck on this horse.

Why not put a single dollar, just as a lottery ticket. If you are the only one, one dollar is enough to collect the jackpot collected for the winner without sharing it. One dollar, You would have a huge return, never reached at the bookies

Richard Burkett

It's hard to imagine that a person would "know for sure" that a horse would lose, barring information that a 'fix' was in. The apparent longest of longshots has been known to come home on occassion.

So I sorta object to the question, unless something illegal/unethical is going on. Acidtest's first answer is therefore the only one to make any sense in 'real life', although several others have interesting angles, but lack the 'know for sure' factor.

Oliver Townshend

You are a bookie (or whoever takes bets in the US), and you want to balance your books.


Let's say that you are bad at picking horses. Whenever you pick a horse to win, the horse loses. Before long, you'd have to switch strategies; instead of picking the winner, you pick the horse that, you think, will most likely lose.


Let's say that .50 has been bet on horse X.

$10 has been bet on horse Y.

Let's assume there is no house take for now.

The pay out on horse X winning is 10.50/.5 = $21.

If you bet an extra $2 on horse Y, the payout on horse X is now $12.50/.5 = $25.

So, you bet $2 extra, but you got paid $4 more.

You can work this out even with a house take, or more horses.


You have horses X and Y.

If less than a dollar has been bet on X, then you can bet on Y to win even if you know he's going to lose.

Work to follow.

Ben K.

The answer is in the breakage, but you need some serious collaboration among bettors to make it work.

Imagine you have exactly 20 people betting on the race. Imagine that all 20, including you, bet $4 on the eventual winner, horse #1. The rest of the betting action sums to 119. So the total betting on the race is $199. If horse #1 wins, the house gets $39.80 (20%) plus $1.20 in breakage (each bettor loses $.06 in breakage.

In this case, the collaborative of bettors would be better served on having one additional dollar placed on another horse (the sure loser). An additional dollar placed on horse #1 would actually increase the house's breakage take to $1.40, but a dollar placed on a non-winner brings the house's breakage take to $0.00.

In theory, each bettor should be willing to pay you up to ~6 cents to encourage you to place the extra $1 bet. You will accept any amount over 5.26 cents per player (from 19 players) in order to place the bet.

The other theory here is that placing a stray dollar on an un-bet horse keeps the house from fixing races so that the un-bet horse wins (and the house collects).


Jared L.

To know a horse is going to win we must assume there is a degree of corruption already taking place. With that knowledge, why not further corrupt the system, right?

Since the payouts depend entirely on existing bets, there is a strong incentive to stack the bets on a losing horse. If we were able to make big bets on a losing horse and also persuade others to make similarly large bets on that horse - we could let an insider know about the winning horse and get paid off when that horse indeed comes through.

As long as the Insider is able to bet more than you personally invest in the losing horse, the outcome will be a net gain (although the bet on the losing horse is a sunk cost).

Avishai Schiff

As long as the bets on the horses do not correspond to the actual chance they have of winning (i.e. a horse with twice as many bets on him than another is not actually twice as likely to win the race), you can make money by “shorting” the front-runner and going long the long shots, no pun intended. If you take a dollar from your friend who wants to be on horse #4 (the horse with $40 in bets on him) and then use that dollar to make a bet on a long-shot, if the front-runner wins, you owe your friend $3, if the long-shot wins, you make almost $12, and if neither wins, you come out even. As long as the front-runner is less than 4 times as likely to win as the long-shot, you will make money.


The answer is "yes." The answer is built right into the blog post. Since the question posed was of a yes/no nature, I win! Right?

Avishai Schiff

The last line of my post should actually say: "As long as the front-runner is less than 4 times as likely to win as the long-shot, you will make money in the long run."


Anjelica Huston did this in "The Grifters".


Clearly, it is because you are betting in bizarro world.