Sports Media, Betting, and College Basketball
As a sports-crazed 10-year-old I was glued to the TV when SportsCenter came on. I, like many other kids from my generation, locked into the iconic sliding bar that filled the left side of the SportsCenter program, displaying what games would be talked about next. The best part of the program for us was waiting until the end of the hour for top plays, a fair reward for the time we spent catching up on the day’s sports news. Now, as a sports-crazed 23-year-old, that child-like wonderment for SportsCenter has since passed. Turn on ESPN at any hour of the day and you will get your favorite ESPN talking heads discussing their takes on the Cowboys, the Knicks, Aaron Rodgers, and LeBron James. For me, that discourse holds very little weight. Because of X, I usually have seen the news story they are showing right after it happened. Missouri upsetting Kansas - saw it as it happened and even saw people on X opine about the quality of court storm. The Mets signing Juan Soto - it was the first thing on my X timeline as soon as it happened. The point I’m getting at here is that the value of sports information for a TV program has never been lower and that has a direct correlation with how sports are being marketed and talked about on ESPN.
The worst part of catching up on yesterday’s sports news via ESPN may not even be the incessant flood of outlandish takes on situations that I don’t care about. Rather, it’s the deluge of sports betting advertisements that promise massive wins to those prospective bettors willing to join their respective sportsbook. What saddens me about this advertising paradigm is that sportsbooks are marketing something they aren’t even selling. It’s well documented that if you are a winning player on a major sportsbook, you will ultimately be limited from placing large wagers on games. So, we now reside in a place where sports coverage preys on the customer’s wallets, gives low value information, and prioritizes individual personalities and their opinions instead of trying to create content that holds insightful, cutting-edge information.
I think that a potential marriage between sports betting information and sports media coverage could begin a new era of meaningful sports information disseminated from the likes of ESPN. Sportsbooks excel at assigning point values to team skill and their line’s “efficiency” correlates with how much money the sportsbook can make. In the same vein, any sports analyst could benefit from seeing a game’s line as actionable information.
Aristotle first coined the idea of wisdom of crowds. This foundational concept says that if you were at a state fair guessing the weight of a cow, the average of the guesses by the collective would be more accurate than that of a single cow expert. The same concept holds true in sports betting. Markets for NFL, NBA, and professional soccer operate as fairly efficient markets because large swaths of money are poured into them. This influx of money holds more validity than any ESPN analyst’s opinion of the game. Today’s sports coverage loves to fire off hot takes with seemingly no objectivity. #1 Kansas losing to Missouri quickly turns into a “Kansas doesn’t have it” story that provides a bear case for their national title prospects. Those analyses hold no grounding in truth and discredit the nuance that goes into evaluating team performance.
Herein lies a new framework for evaluating performances like the aforementioned “#1 Kansas Jayhawks fall to the lowly Missouri Tigers”. Let’s create a fictional basketball tournament that has UW-Madison, UW-Milwaukee, and Marquette face off at the Fiserv Forum. At this event, every team has to play each tournament participant and the lines for the initial two games are already out. The fictional lines look something like this:
UW-Madison -5 against UW-Milwaukee
Marquette -8 against UW-Milwaukee
In this example, Marquette and Madison are yet to play each other but having the two lines above can give us a reasonable idea of what Marquette versus Wisconsin may look like. The market thinks that Madison holds a 5 point advantage over Milwaukee while Marquette holds an 8 point advantage, giving us an estimate of Marquette being a 3 point favorite against Madison.
The above idea comes from Inpredicatable, a website that produces market-derived team rankings for the NBA, NFL, and NCAAF. The founder, Mike Beuoy, was nice enough to outline his system’s methodology years ago and because NCAAB didn’t have an updated system, I decided to create one myself.
My main motivation for creating such a system was what I previously talked about - a disdain for current sports media and the lack of nuanced discourse. Although I have great respect for the likes of KenPom and BartTorvik, citing their lines and statistics as relevant power ratings discounts the power of the market. The beauty of markets is that the line tries to account for all public information. If you are using a public website’s “projected score” to bet against the current Vegas line you will end up being a long-term losing bettor. My power rankings show what the market objectively thinks are the best teams in the country and allows anyone reading the ability to plot two teams against each other to come up with the actual spread of their game.
My code for this project looks at all historical lines and uses KenPom’s home court advantage estimates to create a neutral spread. Decay weights are then added to each game, assigning more weight to recent outcomes/lines than those from the season’s beginning. Once every game has a neutral spread and the data has been cleaned, a regression is run. This regression creates a design matrix, assigning +1’s to winning teams and -1’s to the losing team. If a team didn’t play in that row, it is assigned a 0. While the design matrix holds the winning/losing team values, the target vector holds neutral point spreads for all the games. The linear regression attempts to explain the difference in team skill via a power rating system.
This initial linear regression doesn’t have the most accurate results. Why? We haven’t made any adjustments for what has happened in the game. As fans, we know that the markets will be affected by actual outcomes. Beuoy’s methodology provides a reasonable equation that accounts for these over/underperformances. I won’t get too in the weeds on an example using this equation but you can find his methodology here.
After we have adjusted for spread credibility, we run a second regression. In short, the first regression gives an initial guess of how good teams are based on historical lines, and the second run adjusts those predictions based on what actually happened in the game using Beuoy’s equation.
We now have a comprehensive, objective view of the college basketball hierarchy. The numbers you see in the table below are the spreads of any team against the average team on a neutral court. Please note the final point about it being a neutral court. In college basketball, home court advantage is a sliding scale that can be worth anywhere from 4.5 points to 2 points.
To tie this all together let’s walk through a Saturday college basketball matchup: Marquette at Dayton. In the below rankings, you will find Marquette rated at -20.3 and Dayton at -13.8. On a neutral court, Marquette would be 6.5 point favorites against Dayton. However, Dayton hosts Saturday’s game and KenPom estimates their home court advantage to be worth 3.4 points. By this logic, Marquette should be about a 3 point favorite against Dayton this weekend.
Unlike main-stream sports media, this model has not baked in any bias against your favorite team and that’s the beauty of it. Instead of firing takes on Kansas being an awful basketball team, we can instead use Beuoy’s credibility equation to say that Kansas may just be a few points worse against an average team than we initially thought. The NCAA tournament often gets sold as a high-variance venture which isn’t implicitly a false pretense. However, these market-derived power rankings show that skill disparity at college basketball’s peak is quite small. The small skill gap between top teams provides evidence for how impossibly hard it is to actually win the NCAA tournament. Hypothetically, if a team has an 80% chance to win in round 1, 80% in round 2, and 55% in the four subsequent rounds, their probability of hoisting the ultimate prize would be about 5.8%. Those win probabilities aren’t unrealistic estimates for any top 10 team and expectations should be tempered accordingly.
Although I may portray it as such, markets aren’t necessarily all-knowing. That fact should be the springboard for actual sports analysis. The sports media landscape has a myriad of former players who actually understand the actions and schemes teams are running. And the true beauty of sport, and a core reason why we watch, lies outside of a line. The team’s togetherness, resilience, and intangibles are all things that a model cannot exactly put a number to. As a sports fan, these are the topics I want to see covered by former players who understand those dynamics better than anyone. We are in a new era of sports understanding, we just haven’t fully adjusted to it yet.
If you have any inquiries on this article, please reach out to pipitonevi@gmail.com and if you think these rankings are wrong or biased, I urge you to responsibly bet against them!