The scourge of “win probability” in sports

To watch baseball, or any other sport, is to confront the fundamental unpredictability of the universe, its utter refusal to bend to your desires, no matter how passionate they may be. In recent years, some broadcasters have tried to alleviate this existential uncertainty with statistics. This season, ESPN announced that special graphics will appear on all Major League Baseball broadcasts. In the upper left corner of the screen, just above the score, each team's chance of winning is expressed as a percentage – an integer reassuring in its roundness that is recalculated after each shot. His predictions can help tame the wild and fearful mood of your fandom and limit your idea of ​​what might happen next to a narrow and respectable range.

One might think that reminding fans so forcefully of their team's “probability of winning” would be contrary to ESPN's interests. If your team is down by several runs in the eighth inning, your hopes are already fading. But seeing that sinking feeling portrayed on screen, in clear and precise sound 4 percentmight make an early bedtime more tempting. The producers of reality shows like The Amazing Race They know this, which is why they use quick cuts and split screens to trick fans into thinking the teams are closer than they really are and that the outcome is less certain than it actually is. But ESPN has an evolved consumer in mind. We got a hint of who this person might be in March when Phil Orlins, the company's vice president of production, previewed the graphic. Orlins said the probability of winning would “reflect the way people think about sports right now,” particularly people “who have a bet on the game.”

Sports fandom has always had a quantitative component, but it has become much more pronounced in recent decades. As fans get older, they spend less time watching the games. Maybe they once imitated a favorite player's signature swing or put on a glove and imagined themselves making a game-winning catch. But now that they have left the playground behind them, they no longer identify so naturally with the players. It's easier for them to portray themselves as coaches and business leaders – strategic thinkers surrounded by stacks of Excel printouts. Fantasy leagues were a gateway drug for people who liked their sports with a heavy dose of statistical analysis. Sports gambling apps have become their hero.

As sports gambling became more popular, probability statistics started appearing all over the broadcasts. Apple TV+ has an entire dashboard that sometimes tells you how likely each battle is to end a certain way. Similar graphics pop up whenever NFL coaches consider a two-point conversion. However, these metrics don't seem to be very popular with casual viewers. Judging by the angry fan posts on X, people seem to find them either irritatingly redundant or irritatingly inaccurate. But the graphics have spawned a new breed of postgame memes: When teams pull off an improbable comeback victory, people who might once have taken to social media to share the highlight of a late-inning home run can now share a simple action , which shows the exact moment their team's win probability fluctuated from a low value to a high value. Last Saturday, Reed Garrett, a reliever for the New York Mets, did his best for this practice after the Mets' eighth-inning rally against the Philadelphia Phillies. “Our win probability charts are going viral right now,” he said.

Aside from this niche use case, it's not clear whether these stats are at all helpful to people watching games with the FanDuel app open. When I called Michael Titlebaum, a philosopher at the University of Wisconsin-Madison who studies probability, he told me that these statistics could easily be misinterpreted. “Decades of cognitive science experiments tell us that people are really, really bad at understanding probability percentages,” he said. Even doctors and other professionals who often deal with such numbers routinely make incorrect judgments. Evidence suggests that most players have difficulty converting probability percentages into betting odds and are particularly poor at thinking about multiple such percentages in combination when making parlay bets.

Regardless of its drawbacks, however, the dissemination of probability through culture and entertainment can have positive effects on people's statistical literacy. Kenny Easwaran, a philosopher at UC Irvine, compared it to the way the concept of temperature was valued by the public. In the mid-18th century, some scientists were skeptical that there would ever be a way to represent all the different phenomena of heat and cold—the searing surface of a pan, a steaming jungle, the chill of a glacier—with a single number. But then the thermometer became widespread, and with practice people learned to associate its readings with specific experiences. A similar transition is now underway, he told me, as probability percentages have seeped into mass culture, into weather forecasts, medical forecasts and election reporting.

But the win probabilities that ESPN gives on baseball broadcasts may not be much help because they are generated by a secret proprietary model. ESPN's formula is none in total Black box. The company has proposed calculating live in-game probability from the same types of data streams that other models of this type use. This certainly includes the results of many previous games where scores, innings and runners on base were identical, but the company has not disclosed what all will be taken into account. Is team strength taken into account? What about special advantages on the home pitch, such as stadiums with unusual dimensions and a particularly loud audience? Each fan can continually judge the odds for themselves based on all the games they have seen before and what they personally know about their team. They may have observed a player pinching his back in a previous inning, or they may remember a particular pinch-hitter having unusual success against the other team's closer. Certainly ESPN's model doesn't work at this level. But without knowing the details, you can't really estimate the percentage it generates. It's like looking at an election forecast a week after President Joe Biden's disastrous debate without knowing whether it reflects new polls.

Many viewers would prefer Rawdog games without predictive statistics. After all, anyone who invests in a game is already engrossed – to the point of madness – in the task of intuitively assessing their team's likelihood of victory. Easwaran told me that when there are no numbers, people are actually pretty good at it. He compared it to the organic way we use our reflexes. “If you throw me a ball, I’ll probably catch it,” he said. “But if you tell me it's going to be coming at me at 15 miles per hour and at a 60 degree angle from that particular direction and ask me to calculate where to put my hand, I'm going to feel really bad be there.” If you follow a baseball game closely, you will have noted the score, the inning, and the number of people on base, and these will be reflected in your general state of anxiety. At best, the win probability graph provides a rough quantification of what you are already feeling. At worst, it makes you question your purpose for the game.

That's not to say that sports broadcasters shouldn't have win probability calculators at all, just that the best ones tend to be people who can explain their reasoning. Chick Hearn, the Los Angeles Lakers' longtime play-by-play announcer, did a version of this in the closing minutes of every Lakers victory. He would try to guess the moment when the team would finally eliminate their opponents. “This game is in the fridge,” Hearn said when he felt the game was out of reach, then continued with a refrain that every Laker fan of a certain age can recite: “The door is closed, the “Lights are closed,” out, the eggs are cooling, the butter is hardening, and the Jell-O is jiggly.” Over the years, his refrigerator reputation proved extremely accurate. But occasionally he was wrong, because no matter how good your internal model is, teams sometimes come back against big odds. That's why we watch the games.