NHL’s Latest Player-, Puck-Tracking Efforts Aim To Revolutionize Hockey Production (and Viewing)

Workflows speed the turnaround time and simplify the packaging and presentation of advanced data for broadcast

For Keith Horstman, this is all very familiar.

After more than two decades at the NBA and playing a pivotal role in development of the league’s robust player- and real-time data-tracking system, the NHL’s VP of technology now finds himself at the epicenter of the next generation of data, statistics, and video visualization for the sport of hockey.

Since he joined the NHL last spring, player and puck tracking has occupied the top of his to-do list, and the past couple of months have begun to show the fruits of that labor. After extensive tests, fans got a peek at what the next generation in the hockey-viewing experience could look like during last month’s NHL All-Star Game in San Jose, CA.

The NHL, in association with its enterprise-applications–software and database partner SAP, executed a workflow for major rightsholders NBC Sports and Rogers and the in-venue production team at SAP Arena, streamlining the packaging and display of all this valuable data in a way that was exciting for hockey fans to consume.

Player and puck tracking is here! #HiTechHockey

Get to see this @SAP technology live during tomorrow night's @Honda #NHLAllStar Game at 8:00p ET on @nbc, @CBC, @Sportsnet & @TVASports! pic.twitter.com/uPD8vSSYBM

— NHL (@NHL) January 25, 2019

Using a chip-based system, the league tracks the puck and players on the ice in a more seamless and comprehensive way than ever before. Marrying the new tracking information with its existing in-house Hockey Information Tracking System (HITS), which tallies all the essentials to properly log and score a hockey game (time-on-ice, face-offs won/lost, shots on goal, missed shots, shot types, penalties, etc.), the league and stats and broadcast-graphics partner SMT are simplifying the packaging and visualization of data in producing on-screen visuals that are both compelling to the viewer and easy to integrate for the producer.

Broadcasters dropped in IDs (or pointers) that followed players, trails of players or the puck, and advanced performance analytics at or close to real time. Even NBC Sports Digital streamed out an alternative version of its broadcast dedicated to showcasing this rich data.

Horstman sees few sports — if any — that would benefit more than hockey from the data-tracking revolution in sports, and he believes that the experience at the All-Star Game was just the tip of the iceberg.

“The nature of tracking data really associates with the game more [in hockey] than in a baseball or football game,” he explains. “It goes back to the way we do stats, the old way of putting in single events. Now we’ve got the fluid flow of data for the continuous game. That sets hockey apart from everything else. Not only did the player score from this spot, but how did he get there, or how did the goalie come across the crease to make a save? There’s much more we can do with the fullness of the data in this sport than you can with other sports.”

The Dawn of a Data Revolution?
The NHL’s efforts at All-Star naturally drew the attention of many news entities and garnered some tech-y headlines, but, for the league, it’s about much more than just showing off a few flashy new toys at a tentpole event. This is a system that, when completed, is expected to revolutionize the entire stats and data-visualization process at every level of the league.

In that future, integrating the current HITS system with the new tracking system in an environment where everything from player identification to data distribution is automated will completely change the way the league both documents on-ice play and distributes data to its wide array of partners.

There’s still much work to be done, and the league’s main focus right now is on automating a lot of steps that are currently manual. Today, activating tracking of the puck is a manual process. When the new process is refined, the league will install a series of sensors on the boards near the penalty-box door where referees obtain a puck to be put into play for a face-off. When the new puck passes through that door, the system will activate it and validate data obtained from it for downstream use.

Integration with the clock is also vital and ultimately ties the entire process together. With a tracking system that can essentially log all the actions taking place on the ice by both players and the puck, adding synchronization with the game clock essentially makes the finished product the league’s official comprehensive box score of the game.

“Ultimately, all of the data will join up in a single server and will do all downstream distribution of HITS data and tracking from there,” says Horstman. “We will include an intelligence engine using HITS data, tracking data, and outside algorithms to produce new types of data.”

This would also change the role of the game’s statisticians. No longer will they be loggers of a game’s events but will instead be official auditors of the data being captured.

The league is currently doing a thorough scan at each of its arenas to establish exact coordinates of the ice surface, establishing a foundation for the player- and puck-tracking layer of the system. According to Horstman, about 75% of the arenas have been scanned, with the rest to be completed in the coming month. After the scan, a site survey is done of the arena, and installation of the system can begin. The goal is to have all 31 NHL arenas outfitted with the system prior to the start of the 2019-20 season in the fall.

According to Horstman, hockey is perhaps the most challenging of the major team sports within which to establish such a system. Besides the game’s speed and close quarters, the players are much harder to identify than in a sport like basketball, where everything from skin color to hair color to tattoos and sneakers help individual players stand out. In hockey, there’s basically the name and number on the back of the jersey, and that’s it. Incorrect identification of a player could have dire consequences, throwing off everything else in the process.

“You [would be] sending incorrect data downstream to betting providers, all of your stats systems, and everything else,” Horstman points out. “It is a different, much more difficult problem [than at the NBA]. Having gone with a chip-based tracking technology has alleviated a lot of those concerns.”

Working Out the Kinks
In the months leading up to the All-Star Game, the NHL undertook some key internal tests, including an in-depth session at a Las Vegas Golden Knights game in early January while the Consumer Electronics Show (CES) was in town.

No work left the building, but the NHL and its various partners were able to test out services, products, and the overall workflow of the holistic tracking and data-distribution environment. Basically, how (and would) this all work during a real game?

“It’s not used in regular-season games for a reason: we’re still tweaking, manipulating, and trying to optimize,” says Dave Lehanski, SVP, business development and innovation, NHL. “What is so encouraging, even in just the past couple years, is just how many more opportunities there are to use the data to grow the game, to help and enhance coaching. What we can do from even just a digital standpoint in the development of new apps — be they AR or coaching or live sports betting. The opportunities to use the data are endless right now.”

During tests in Las Vegas in early January, the NHL’s tracking system was able to record player and puck motion in real time and could also re-create game action in a virtual environment using that information.

He also says that much of the effort has been helped along by external developments, including the opening of legalization of sports betting in the U.S. The data, along with the optimal automation of that data, is of huge value to companies like Genius Sports and Swish Analytics (which have worked alongside the NHL during testing) in building live betting experiences using the downstream funneling of the data.

“We are the beneficiaries of things happening in the world outside of our control,” says Lehanski. “The way that sports-gambling practices and policies have evolved in the U.S. has changed that entire business and created a new opportunity for what we can do with the data. That makes it more encouraging and exciting than maybe it’s ever been to see that we’re really close to making it a reality.”

Perhaps one of the most interesting lessons learned from the buildup to the All-Star Game was that, among all the players on the ice, the toughest to track were the goaltenders. The players doing the least amount of traveling around the ice required the most TLC in the chip-positioning process.

“It was very surprising,” laughs Horstman, “that, over the couple weeks leading in, we were having the most difficulty with the goalies, who were just standing facing the ice surface.”

Originally, the goaltenders’ chips were slotted into a pouch sewn into the back of the neck of the jersey. In that position on the body, only a couple of the receivers in the rafters were able to receive their signal, making the goaltenders’ data much less refined than their open-ice counterparts’. Eventually, the NHL was able to identify a spot in the chest protector where the chip could be moved to the front of the player and remain safe from incoming pucks.

Simplifying Data Packaging for Broadcast
The NHL is hoping to differentiate its tracking efforts from other leagues’ by making visualization of all this data in the television broadcast a part of the process from the beginning. Tracking and real-time data rapidly loses its value if not implemented and shared with the fan and viewer in as close to real time as possible.

The advanced-graphics experts at SMT have played a massive role in helping the league accomplish just that by making the products of the league’s new system both palatable to fans and easy to integrate into broadcast workflows.

“A lot of leagues that install a tracking system don’t think about how they can package it to make it useful for those entities that might want to use it,” says Gerard Hall, founder/CEO, SMT. “Putting a circle around a player alone is not enough to prove this is valuable. I applaud the NHL — and [Commissioner] Gary Bettman’s vision — on this; this not about just collecting some tracking data and telling [broadcasters] to figure out the rest of it on your own. The NHL is really interested in packaging this and listening to what the broadcasters want, and they’ve hired the experts in the world in this (i.e., SMT) to help that process.”

SMT has developed the next level of HITS with what it is calling the OASIS system, which acts as a meeting point for everything in the chain: tracking data, HITS, the official clock and score. This central repository becomes the point at which the NHL disseminates all this data to multiple APIs, real-time push data feeds, and any other NHL constituents interested in receiving the data.

A key role of OASIS is to serve as a layer beneath everything the broadcaster does in its traditional display-graphics workflow. OASIS uses artificial intelligence to calculate in real time all the events taking place on the ice. For example, it can infer the players taking a face-off based on their behavior during a stoppage of play in the face-off circle. That means that a graphics engineer can automatically identify the players taking a face-off and, with a single click, identify them with on-screen pointers.

“In live TV, you can’t spend time doing all of that manually because you’ll never get it on-air,” says Hall, noting that even player and puck trails can be quickly dropped in live when integrated with the game’s standard graphics package. “The idea is that, underneath it all, there is an artificial-intelligence engine supporting broadcast [and] packaging things up. That’s extremely valuable and makes this stuff practical in a live environment.”

OASIS also can help simplify the packaged-replay process, including dramatically decreasing the turnaround time on packages that feature tracking data and analytics processed by the entire system. According to Hall, the marriage of all the data in the system enables OASIS to place context around an event. For example, when the intelligence engine detects that a goal has been scored, it can do simple tasks like identify the shooter, the passer, the trail of the puck, and the time the puck entered the zone. However, it can also use all that information to determine whether it was a breakaway goal or one that featured a lot of passing in the zone, thus giving the goal context.

The context of the goal helps triggers one of numerous prebuilt customized replays that can drop in all the essential information the producer wants with, essentially, one click. Because a graphics operator doesn’t have to custom-build the tracking graphics for replay, the time required for the replay to be used in a broadcast is reduced.

“In my view, it’s going to change the way TV produces events, particularly in hockey,” says Hall. “It exploits and leverages real-time tracking that goes well beyond just putting a pointer on somebody. It’s revamping production workflows for the NHL in the future, and we’re just at the start of it.”

Going forward into next year, the NHL’s goal is to offer this level of data packaging for all its  broadcasters at the national and regional level, making it a part of all game broadcasts night in and night out and not a flashy toy reserved for big nights like All-Star or the Stanley Cup Final.