Real-Time Win Probability Will Change Football Forever
Machine learning will create a fan-centric statistical revolution.
People love sports for being unpredictable, but that doesn’t mean sports are actually unpredictable. It just means they feel that way. And no one knows this better than Bobby Skoff because Bobby Skoff, the Co-Founder of Swish Analytics, has an automated machine learning platform. His machine sucks in statistics, runs them through a set of algorithms, and pumps out player level predictions for every major U.S. sport. Does he know what’s going to happen before it happens? Of course not. He’s just a lot better at guessing than the play-by-play guy.
Among all the statistics his platform can produce, the one that stands out for Skoff is win probability, the percentage chance that a given team can eke out a victory at any given point in a contest. He’s an evangelist for this information because it allows fans to better understand what is happening in any game, but particularly in football games, which are governed by arcane rules that experts often struggle to explain (just ask a cheesehead what constitutes a reception). To help out this season, Swish has released a free app called Live A.I. that provides fans with live win probabilities for NFL games.
The app won’t change how football is played — at least not in the short term — but it could change how the sport is consumed. And thats no small thing when over 200,000,000 people are tuning in every season.
Skoff and his founding team, who previously built machine learning software for financial companies like Capital One, are realistic about Live A.I.’s impact on fans during the 2016-2017 season. The app has over 10,000 users, but critical mass is a ways off. His plans, however, are ambitious. He wants to help fans know when to get loud, know which bets to take, and know when to start drowning their sorrows.
Inverse spoke to Skoff about being the best informed fan in the bar.
You had success working in financial services. Why take your skill set, machine learning, and apply it to sports? The commentariat there is already massive and it’s a tough arena in which to fight for relevance.
You can go on Twitter right now and find a hundred guys that are going to say, ‘Hey, my past 100 NFL picks did perfectly, sign up for my product for 500 bucks right now’ and it’s shady as hell. We took a more Silicon Valley approach to a world that operates in the dark. We show the result of every prediction we’ve ever made, we slice it and dice it to allow you to better ingest it, better understand where we’re doing well, and where we’re not. It’s a first of this kind of experience in the sports world.
When did you first realize that your machine learning platform had real potential as more than a personal project?
We broke off pieces of historical NBA data originally, and had success retroactively predicting games. We found ourselves saying, ‘Hey, if we would have bet in all our predictions, we would have won.’
Live A.I., your app, surfaces win probability information in an idiot-proof way. Its so easy to use that I’m wondering if you’ve thought about it becoming an alternative to traditional scoreboards.
I think of Live A.I. as having an educational component, especially for Europeans, because football has so many different rules. The game has visually insignificant plays that actually, from a probability standpoint, have really big impacts on the game. Probability allows you to see the most impactful play of the game.
And for people who already understand the sport?
On NFL Sundays, you have all these games going on with RedZone and all these other channels. You usually have to pick between a couple different things to watch. Probability allows you to change the channel when a game is statistically over. On the East Coast, Sunday night, Monday night, and Thursday night games run well past midnight, which is just insane. Win probability can be close to 100% in the third quarter. At that point, turn the lights out, turn the TV off. It’s over.
I actually had my fiancé at this 49ers game a few weeks ago. In the third quarter, the Cowboys are winning and the probability goes toward 100%. She sends me a screenshot and I’m like, ‘Beat traffic, you might as well leave. You’re not going to miss anything. The game is done.’
I think the 49ers can be described as an ‘edge case’ at this point. How can the NFL or broadcasters leverage the information you’re presenting? Is your platform going to change the way I watch games, even if I’m not engaged with your app?
We are working with some companies around real-time advertising. How can you better either logically serve ads to customers based on this information, or can you present them a really engaging ad because of what’s going on? Hey, it’s the end of the third quarter, you’re a fan of the Knicks, they have a 0% chance of winning, you need a Bud Lite.
The creative around that ad, and kind of the flavor of the ad you get, would definitely be driven by the win probability. In some pilots. It’s like, ‘Hey, your team is about to lose, sob your tears into a Whopper. Here’s a coupon for $2 off!’
But you have to be open to the fact that improbable things do happen, right?
The Chargers had more than 85% chance of winning games with less than five minutes to go for four weeks in a row. Somehow, they lost. That’s pretty crazy.
**Those must be some crazy looking line graphs. Do you think seeing the narrative arc of games visually — and having fans see that — will potentially highlight some of the structural issues with specific sports?
Once the game is done and you see the whole thing drawn out, you get a very clear understanding of what went down. Because of the nature of the NBA, if it’s within ten points, every team is in it and it’s all a pre-game thing until the fourth quarter hits. Obviously, there are lots of exceptions, but people will notice that sort of thing.
You think the ubiquity of win probability is inevitable, don’t you?
I think it’s something that should be on the broadcast, and if they’re not going to have the guts to put it there, then it belongs on a second screen.