FOLLOW

FOLLOW

SHARE

Will Machine Learning Transform Sports Viewing?

AI could tell you exactly how likely your team is to win

1Nov

The principle reason that sports fans enjoy watching games is their inherent unpredictability. Ahead of admiring physical ability and skill, a sense of loyalty to the team, or the match day hot dog, the unpredictability of sport is what keeps millions of people tuning in and passing through the turnstiles every week the world over. It’s because Buster Douglas can knock out Mike Tyson, the Jets can beat the Colts in Super Bowl III, and Leicester City can win the Premier League title. For all you can say that a team or individual is ‘likely’ to win, you can never be absolutely sure.

So sport is certainly unpredictable. But random? Absolutely not. The sports industry is one of the most closely scrutinized and publicly examined in the world, with millions of data points being collected each week, almost all of them being available for the viewing public to pick apart. So much goes into analysis of the games people love that the job of betting companies is only becoming ostensibly more and more difficult, as customers substitute gut instinct for cold hard numbers.

It’s onto this data-heavy landscape that companies like Swish Analytics are developing software. The company’s free app Live A.I. provides users with real-time win probabilities for NFL games. Each play is presented alongside its effect on the win probability of the team that made it, and the company says it has a pretty good success rate in predicting the outcome of games.

‘We broke off pieces of historical NBA data originally, and had success retroactively predicting games,’ Swish co-founder Bobby Skoff told Inverse. ‘We found ourselves saying, ‘Hey, if we would have bet in all our predictions, we would have won.’’ A lot of people offer an analytical approach to sports betting for a fee but, as Skoff says, these are often ‘shady as hell’ offers promoted on the likes of Twitter and other less official channels, and the industry is ripe for a more professional offering of win probability. Importantly, Swish doesn’t deal in odds, it presents a percentage change of each team winning the game - this isn’t just for gamblers, it’s for the more casual fan too.

In his interview, Skoff also suggested that the win probability calculator could be used to notify spectators as to when a game ceases to be a genuine competition, i.e. when the probability of one team winning edges close to 100%. Skoff suggested that fans could leave games early to beat the traffic, or switch off the TV and focus their attention elsewhere.

This, though, is a particularly US-centric suggestion - in Europe, in particular, there is less of an obsession with finding out who finishes a competition on top. A prevailing US issue with soccer, for example, is the notion that two teams can, and often do, share the spoils at the end of a game; American sport doesn’t really cater for this. You would be hard pressed to find a European fan that would leave a stadium in which their team had gone far enough ahead to have killed the game - they revel in the victory. And, conversely, fans of a team that is clearly losing will, up to a point, stay and support their side to the death.

Outside of the US, then, it’s the game-enriching applications of the technology that will be the most appealing. The play-by-play analysis, complete with each play’s impact on the overall game, will find drama in otherwise ordinary periods of the game, helping even the most casual football fan enjoy the less explosive periods. The casual gambler, too, will be more up to date than ever with the likelihood of a particular outcome, and in-play betting will doubtless see a rise. Machine learning solution like Swish will keep popping up as the technology develops, and the experience of viewing will change as they do. Swish won’t have soccer fans leaving stadiums after 60 minutes, but it might help fans understand the sports they love. 

Comments

comments powered byDisqus
Social media small

Read next:

'Structure Is The Next Big Thing In Social Media Analytics'

i