I always like to say that horseracing was my introduction to markets. It’s not exactly the same, but as a young, curious kid growing up in Kentucky, I do think it planted the right seeds for a later interest in investing. (To my current clients, do not worry: my only gambling takes place during the first weekend in May.) Both Seth Klarman and Warren Buffett reportedly had similar interests growing up—it’s a fun puzzle, just like markets.

While there’s not much research on the subject, what has been written shows that the betting markets at horsetracks have their own fair share of inefficiencies. Here’s an interesting one to think about for anyone looking for an “edge” at the Derby this year, although this is certainly not financial advice:

Not everyone realizes that the horsetrack betting system is in fact a market. Unlike other sporting events, the odds are not set by a central bookmaker. Tracks instead use a parimutuel system. The track receives wagers for different betting pools for each race, and winners are given their proportional share of the losers’ money after the track takes its cut from the pool. Instead of betting against the house, patrons are betting against each other. The odds seen on the toteboards are updated each minute based on the total wagers received for each horse.

This has a few implications. First, as a bettor, you are not betting against a house with considerably more resources than you, and you are instead relying on the crowd to set fair odds (important). Second, the odds & payout of your wager are not set in stone until after the betting windows have closed for the race (also important, but save it for later).

Let’s start with that first idea, that the crowd is the one setting the odds. There’s a famous experiment where a large group of individuals had the task of individually estimating the number of jellybeans in a large jar. So long as the crowd is large enough, the average of their estimates will typically very closely approximate the actual value. When the crowd is smaller, their average estimate is less accurate. As a consequence, we can say betting pools with more participation are a better reflection of the true odds of the race.

We can also take the perspective of a very successful bettor, who has some skill in making wagers and has lots of money to throw around. As they bet, the market odds shift closer to what should be a more accurate representation of the true odds. Under a parimutuel system, they would prefer to wager in larger pools such that they could extract more money from the track. Smaller pools may not provide the same liquidity needed to place large bets without drastically skewing the odds. We should avoid this bettor, since they likely know better than we do.

With these two ideas, we can come to the following conclusion: less popular betting pools likely have less competition and have more opportunities for smaller, enterprising bettors. More popular betting pools, by contrast, more accurately reflect the true odds of the race and are less likely to have attractive opportunities. In this case, we should clearly prefer to be betting in pools that are less popular amongst the track goers and sophisticated players. This is game selection: skewing the odds in your favor by choosing the game that yields you an advantage over other players. Buffett had a nice way of putting it with a poker analogy:

“If you’ve been playing poker for half an hour and you still don’t know who the patsy is, you’re the patsy.”

The illustrative strategy I am sharing with you here relies on this idea. It won’t make you any better than the crowd at predicting winners, but it may help you think about finding some mispriced bets otherwise. 

Placing a “Win” bet at the track is simple to understand: if the horse you bet on comes in 1st place, you’ll win money. The amount you’ll win is easy to calculate based on the odds of the horse shown on the toteboard. Consequently, this is a very popular type of bet to make. On the other hand, placing a “Show” bet is slightly more complicated, since you will be paid if your horse finishes in 1st, 2nd, or 3rd place, but the payout amount is uncertain. You still get your share of the money wagered on the horses that do not finish in the top 3, but this amount is split amongst the three horses that do finish in the top 3. As a result, show bets are significantly less popular. 

The clear thing to do from a game selection standpoint is to look to the show pool for opportunities. And sometimes, there are large discrepancies between the proportion of money wagered in both pools for the same horse. Here’s an example from April 11, at Keeneland in Lexington, KY:

Race 1

	Horse	$2 Show Bet Payout	Win Pool %	Show Pool %
1st	West Coast Dream	$4.04	~19%	~13%
2nd	Speed Skater	$3.54	~18%	~16%
3rd	Lexico	$4.06	~12%	~13%

So in this race, West Coast Dream was the favorite in the “Win” pool and ended up finishing in first place. We can tell that West Coast Dream was the favorite because it had the largest share of money wagered for it to win. In this case, bettors correctly assumed that West Coast Dream would run the race the fastest.

However, the same horse was not the favorite in the “Show” pool. As you can see from the table in the last column, West Coast Dream only commanded a 13% share of wagered dollars. Instead, the horse with the most money wagered in the Show pool was Speed Skater. 

As a result, a Show bet actually paid significantly more for the favored West Coast Dream than the less favored Speed Skater, and paid approximately the same as the much less favored Lexico. There was a very large gap between the market expectations in the Win and Show pools. That would not make much sense in an efficient market! 

The idea is simple. If we see that there is a wide gap between the proportion of money bet in the Win pool on a horse versus what is bet in the Show pool, there’s probably value in that bet. We should expect that the Win pool is generally more efficient, and use that to inform our expectations of the race. In efficiently priced races, the proportions between the two should usually be approximately the same. The odds of having a winning Show bet are higher, but the payout is proportionately lower.

So, what’s the catch? There are at least two that I’ve come up with. The first is that the track has a “take-out” of anywhere from 10-20%. This is money removed from the pool to compensate the track. This can be accounted for by only taking bets where the difference in odds is sufficiently large enough to cover the take-out. 

The second obstacle is more difficult to account for: we don’t actually know the final odds in both pools until the windows are closed. We can keep track of both up until the final moments, but since the odds are ever moving, we cannot know for certain that the discrepancy is not also being corrected by someone else. See, if someone else has the same idea as us, and we both choose to bet at the closing moments, we may both end up overbetting and skewing the odds in the other direction. 

For this to work, we need to count on three things: (i) the odds shown for winning are accurately set by the crowd, (ii) there needs to be a large gap between the Win pool % and Show pool % such that the Show bet looks cheap, and (iii) we need to be sure nobody else will also be looking for this opportunity. So, while this inefficiency is real and can be observed, it’s not as simple to take advantage of it. I’m still figuring that one out, and that’s what makes these kinds of game theory puzzles so interesting. And—one more thing—I’m not giving out any kind of secret here. This anomaly has already been documented by researchers and has been around for quite some time (Hausch et al., 1981).¹ You can read their paper for the math if you’d like.

What Does Any of This Have to Do With Investing?

As I’ve alluded to, financial markets and parimutuels aren’t so different. (However, the track is not a good place to look for investments. Don’t confuse the two.) When you’re picking stocks, it’s important to find areas where you have a leg up on the competition of other investors and analysts. At Appalaches Capital, I have two areas I like to focus my efforts in. The first is finding value in stocks that may require a long-term view. The second is finding value in stocks that others may wish to avoid for non-fundamental reasons.

Most of the investment industry is oriented around monthly, weekly, or even daily reporting. Managers are evaluated on the short-term fluctuations in their portfolio performance, and as a result are keen to look for what stocks are “working now” or have immediate catalysts. So, if you have a fairly boring stock without much current excitement, such as the AutoZones of the world, it’s reasonable to believe that not many others care to reach for it. Or, maybe a company has a weak current outlook, but it seems likely that it will change in the coming year after some adjustments are made. Investors with shorter attention spans may wish to wait for the upswing, leaving value for the patient.

Still, even among longer-term oriented investors, many shun certain stocks at any given point in time for a variety of reasons that have little to do with the company’s performance. One example is a stock’s correlation with the S&P 500, or some other broad benchmark. Many active managers wish to track the benchmark somewhat closely, and buying stocks that do not move in concert with the market increases the risk of failing to achieve this goal. If you’re willing to step out of line, this isn’t a bad place to look either.

Now, during this Derby Weekend, you can ponder all of the different dynamics going on within the betting market and where people are making mistakes. Or, while you’re watching the greatest two minutes in sports, you can just root for the horses with the funniest names. That might be more enjoyable.