Monday, July 23, 2018
Takeout: "The Price You Pay"

Now they'd come so far and they'd waited so long
Just to end up caught in a dream where everything goes wrong
Where the dark of night holds back the light of day
And you've gotta stand and fight for the price you pay
Oh, the price you pay, oh, the price you pay
Now you can't walk away from the price you pay
~ Bruce Springsteen

Un_dollar_usCall it what you will... the vig, the rake, the juice. In the world of pari-mutuel gambling it is commonly known as takeout. By any name it is the invisible hand that reaches into your pocket -- the cost associated with making a wager. For you, the sports bettor, the horseplayer, the slots player, it is "the price you pay".

Takeout is a fundamental necessity in the economics of Thoroughbred horse racing. Bettors fuel the pari-mutuel pools and a percentage is deducted to fund purses, pay taxes and expenses, and generate profits for the racetrack. Without betting, racing ceases to exist as we know it. Pat Robertson, the media mogul, moral crusader, and frustrated Thoroughbred racing enthusiast and owner, once said "I don't bet on my own horses and I don't think anyone else should either." Well, Pat didn't quite understand the economic model that allowed him to race his horses for prize money. But you probably do.

The issue then is not the existence of takeout but the rate of takeout. Across the landscape of Thoroughbred racing the effect of potentially excessive takeout rates has become an increasingly debated and thorny issue. For the best (and to my knowledge the only) comprehensive resource on current takeout rates see this chart from HANA.

It is well known that the powers-that-be in California opted to raise certain pari-mutuel takeout rates on Thoroughbred racing in their state. Santa Anita president George Haines defended the price increase when he stated "Maybe the top point-one percent of the handicapping world has that [takeout] in their equation..." This is quite an interesting statement. Firstly, point-one percent of anything is a minuscule proportion. The phrase "in their equation" catches our eye. Now that's something we can sink our investigative teeth into. What factors are included in a bettor's equation? And by equation we mean multiple considerations, i.e., the evaluation of different factors. To what degree do bettors include takeout rates in their decision process?

take imp issueTakeout rates are just one of the myriad of tough issues confronting Thoroughbred racing. If you think that takeout is an issue of utmost importance, you probably pour a healthy amount of money into the pari-mutuel pools. This makes perfect sense. Even a percentage point or two difference can be felt in your wallet.

And if you think that takeout is an issue of utmost importance, you might also think the industry is suffering due in large part to high takeout.

In our recent survey of Thoroughbred racing fans we asked respondents to rank order the importance of selected issues. This chart depicts the selected most important issue segmented by bettor type (this segmentation is discussed in detail later) and shows some substantial differences. Heavy-Hitters (high volume and professional horseplayers) are much more likely than Recreational bettors to consider takeout rates most important (56% vs. 6%). And Heavy-Hitters are twice as likely as Hobbyists to consider takeout rates most important (56% vs. 25%).

A special section elsewhere in our survey focused on the variables a bettor might consider within the decision of which races to play. The analytical technique used is known as Self-Explicated Method (SEM), a simple but robust alternative to Conjoint Analysis (CA). Both SEM and CA are research techniques that attempt to "read the mind" of the survey respondent and reveal their preferences for which features a new product or service should have and how it should be priced. This popular approach attempts to simulate "real world" decision making and the trade-offs that a consumer makes as part of the mental "buy/not buy" decision.

In our case, we've adapted SEM to explore and better understand the race selection process from a horseplayer's viewpoint. We asked respondents (bettors) to express their degree of preference for various levels of four features (variables) that theoretically influence race selection -- field size, purse level, racing surface, and takeout rate. And to allow for a more specific examination of takeout rates we asked respondents to specifically consider an exacta wager.

This is, of course, an overly simplistic model of a bettor's decision making process. Pool size and other less easily quantified factors -- familiarity with jockeys, trainers, and horses for example, surely play an important role for many bettors. And more sophisticated horseplayers are looking for situational opportunities based on trusted angles that may trump other considerations. But as an investigational exercise these four primary factors under evaluation can still yield insights and generate hypotheses for more focused future research.

To gain more insights into this SEM analysis we segmented respondents according to betting volume. This behavioral/usage segmentation is simply defined by a self-described bettor typology question from our survey. After necessary filtering, the following analyses are based on three meaningful bettor segments: "Recreational" (n=76), "Hobbyist" (n=111), and "Heavy-Hitter" (n=39). For an analysis like this we would ideally want larger samples for each segment, but as we will see even these results pass the "sniff test" for face validity. A fourth segment, albeit poorly understood and deserving of special study in its own right, are the "Non-Bettors". This overwhelmingly female segment is by definition not relevant to this wagering analysis.

take field sizeThe first feature that we will look at is field size. The iconic horseplayer and author Andy Beyer suggests that chronic small fields are the biggest reason for Thoroughbred racing's handle decline, even more harmful than high takeout rates and drug issues. Here we get a glimpse into how bettors might regard the importance of field size.

Note the "part worth" values, or utility scores, plotted along the Y-axis. The way to interpret these values is straightforward -- higher values represent more preferred levels of the feature than lower values. The part worth value represents the worth of any given part of the product. In our case, the product is a race.

No surprise here in that small fields are disliked no matter how much one tends to wager. It is interesting that for Recreational Bettors and Hobbyists field sizes of 9 or more bring no higher utility, while for Heavy-Hitters it's "the more the merrier". We can speculate that perhaps there can be "too many" entries for some low-volume players -- a handicapping burden of sorts.

take purse levelWhat about purse level? In our model purse level is understood to be an imperfect surrogate for "quality" of a race. Is it a myth that bettors prefer races with higher quality horses, i.e., faster, classier competitors? As our data suggest, the answer is "it depends".

Given a choice virtually no one prefers to wager on the cheapest races. And both Recreational and Hobbyist bettors tend to prefer higher quality racing.

But the story is markedly different for Heavy-Hitters. Never mind that even the most hard-boiled horseplayer is fervently trying to dope out this year's Kentucky Derby winner -- the biggest players don't care much about quality (purse level). They are likely easily satisfied wagering on claimers of average ability so long as other more important product (race) features are present.

take surfaceAnother race characteristic used in our SEM analysis was Racing Surface. The synthetic racing surface controversy is still simmering and as these data illustrate, if forced to choose bettors of all kinds still prefer natural dirt and grass.

We won't say much more about this one since our primary focus is elsewhere. This issue though is complex with no true consensus as yet on the connection between equine safety and synthetic racing surfaces. 

Still, one has to wonder if the two successful "lifestyle" racetracks that run on synthetic -- Keeneland and Del Mar -- might see a significant boost to their already healthy handles with a dirt track instead of synthetic.

take takeout
hat brings us to Takeout Rates. Recall that respondents were asked to state preference among various takeout rates for a hypothetical exacta wager. The range of takeout rates offered as levels is realistic -- these rates represent the current general range of exacta takeout at North American racetracks.

The data depicted in the chart is exactly what you would expect to see. All bettors prefer lower takeout, and the lower the better. For Heavy-Hitters, takeout is very important and "worth a lot" in their race selection decision.

And it should be no surprise that Heavy-Hitters are the most price sensitive segment. What is a little surprising though is that even Recreational bettors seem to "get" takeout. Their preference for different levels of takeout is not random. While it is not an important part of their equation, i.e., the part worths are low, if forced to choose they prefer lower takeout. And Hobbyists are predictably not as price sensitive as Heavy-Hitters.

take imp factorsHaving stepped through in detail the race characteristics of field size, purse level, racing surface, and takeout rate by looking at part worth values (strength of preference), the next summational chart essentially depicts what our four bettor segments' "handicapping equation" might look like.

What factors drive the decision to play an exacta or pass on a given race? As you already know by now, it depends. Field size is very important to everyone. Andy Beyer is right that small fields result in an unappealing product. But field size isn't everything. In general, the more financially involved a bettor becomes the more price sensitive they become -- they care about takeout rates. And I would argue that high volume horseplayers and dedicated hobbyists have never been more educated and informed with regard to takeout as they are today.

One can easily appreciate the strength of this kind of trade-off analysis in this context. It is easy to imagine a more sophisticated research study with a much larger sample that could include product features that do not currently exist. In other words, we could have easily tested lower takeout rates than those currently offered in an attempt to find preferred combinations of features and more optimal pricing.

The concept of optimal pari-mutuel takeout rates has not received enough attention. The underlying theory is easy to grasp -- there is a perfect price point that maximizes revenue. The rate is not so high that price sensitive players are scared off. It is not so low that the racetrack can't pay its bills. As in other forms of gambling, the rate is set to give back the optimal amount to players who will churn their return back through the system at a predictable rate.

Some will argue that lower takeout rates put more money back into bettors' pockets so they are not tapped out too quickly. Presumably losing your money more gradually renders more enjoyment and increases the likelihood of a "repeat purchase". We can all agree that the casinos have figured this out.

Others argue that serious gamblers in other pursuits like sports betting and poker would give horse racing a fresh look if takeout rates were decreased. Rest assured there would be no stampede if you walked into the slots room at a racino and announced that the track had lowered takeout to 12% across the board. You might get one or two quizzical looks though.

Just as takeout rates have arbitrarily crept upwards over the years, how wise would it be to simply reduce them in a similarly capricious manner?

In the Thoroughbred racing industry takeout rates are set mostly by statutory process with little to no underlying science guiding the way. To be sure, questions swirl around the concept of an optimal takeout rate in the age of full-card simulcasting and phone/internet betting. Does it vary by pool, i.e., wager type? Does it vary by jurisdiction? Does it vary by type of bettor, i.e., on-track/off-track?

The industry has not been convinced that lowering takeout is in its best interest. Even when a racetrack or jurisdiction promotes a special bet with lower take it typically describes the offering as "giving something back to the players", a concession. This is a far cry from doing so believing it will boost the bottom line.

Getting there will not be easy. Historical studies from the literature are of limited value. Fresh, ambitious research is needed -- framed by the modern pari-mutuel system and driven by modern analytical techniques. Although the logistical challenge of natural experiments is daunting, artificial simulation offers hope. Not that such a project should be undertaken without considerable planning and careful thought. Requirements would likely include a cross-disciplinary team of economists, researchers, and industry experts. As well as unfettered access to required data. Not to mention the assistance and cooperation from one or more advance deposit wagering (ADW) firms.

But if not now, when? Few who currently make their living in Thoroughbred racing have any time for visionary thinking as the day-to-day grind is all-encompassing. But this really isn't visionary thinking anyway. It's just smart business and should have been a priority decades ago.

Unlike other tough issues on racing's plate these days, the quest for scientifically-derived pari-mutuel takeout rates should not be divisive and controversial. In the end, if successful, everybody wins.