The NHL’s launch of a new faceoff probability stat is another example of how technology is changing how we watch sports. The new technology uses data and in-game analytics from the past decade to predict the chances each player has of winning the circle draw.
The league has teamed up with Amazon Web Services (AWS) to make this all possible, unveiling the new feature on March 1, 2022. The faceoff probability status promises to add another layer of excitement to an already thrilling sport.
Predicting the Outcome
Whether at the beginning or end of the game, watching a faceoff is one of the most exciting moments in an NHL game. It determines who will have possession of the puck and ultimately could mean the difference between winning or losing. Since betting is a big part of the NHL experience, spectators are always looking for ways to predict who will win a particular game. Checking the NHL odds is one way to determine how various teams stack up against each other. The NHL is giving fans another way to make predictions during a game with its new faceoff probability stats. The technology determines the likelihood of each player in the circle winning the draw. Analysts expect the stat to be used for real-time in-game betting in the future.
How Will it Work?
During a press conference about the faceoff probability stat, the president of the NHL explained that since hockey is played as a series of several events leading to an outcome, the league has identified the faceoff as a critical component of the action. The odds of a player winning a faceoff and the puck’s possession will be displayed live during games on the screen for fans watching game broadcasts.
The NHL’s faceoff probability uses a machine learning model to predict where a faceoff will happen on the ice. The same technology determines the likelihood of either player winning the face and uses that data to predict the winner.
Whenever play stops during a game, the faceoff probability model crunches numbers to generate win probabilities and the players who will go head to head in the next faceoff. Player metrics, faceoff location, and the current game (score, etc.) are used to determine the probabilities.
These AWS-powered machine learning stats are a first for NHL Edge IQ. Hockey is unpredictable, with players performing in surprising ways throughout the season. So, it’ll be interesting to see how accurate the probabilities turn out to be.
Where Does the Data Come From?
The NHL and AWS have used ten years’ worth of data from the Hockey Information Tracking System (HITS) and the NHL’s Puck and Player Tracking data to build the machine learning model used in the faceoff probability stat.
The data is complex and includes things like a player’s height, weight, handedness, head-to-head matchup history, home and away faceoff statistics, and the game’s score. The faceoff probability stat uses all of this historical data plus the current game situation and game context to determine what will happen during the faceoff.
Given the NHL’s popularity, the league is always looking for new ways to engage its fans. Since the faceoff is often one of the most exciting and contested parts of a hockey game, this new technology promises to add to the thrill of watching this part of the game. If you’ve ever attended a hockey game, you know how much tension builds towards the end of a close game leading up to a faceoff.
During a faceoff, fans wait on the edge of their seats for the puck to drop, knowing whoever wins possession can change the game’s direction. With face-to-face probability, fans will see in real-time which player has the best chance of taking control of the puck and possibly the game.
Main Image Credit: