Inside NHL.com’s revolutionary new features, stats
Imagine hockey stats that span a century, allowing you to compare an NHL player’s offensive output in 2015 with that of an NHL player from 1925.
Imagine having the ability to search NHL stats to find something as specific as the number of times a 6-foot Latvian center has scored a game-tying goal on a Tuesday.
Imagine the perfect marriage of state-of-the-art player tracking technology – putting a chip in the puck and on the players to collect data – and real-time stats from a game that redefine concepts like “puck possession.”
And imagine, using that player-tracking tech, being immersed in a 3-D digital recreation of an NHL game, allowing you to swoop through the rink like it’s a video game. It’s one thing to follow the puck; it’s another to follow the entire game from the puck’s POV.
Hockey fans: Prepare to tumble down the Internet rabbit hole.
The NHL and SAP, the multinational software corporation that owns the naming rights the San Jose Sharks’ arena, have partnered on revamping NHL.com’s stats pages in the short term, and integrate player tracking data into the site in the long-term.
The revolution begins on Feb. 20, as the switch is flipped on the new NHL.com stats pages. Here’s what to expect from the League over the next two years, as it revolutionizes the way we watch and analyze the game.
PHASE ONE (Feb. 2015)
The NHL and SAP were in negotiations for around nine months to bring to hockey the sort of innovation the company had already handed to the NBA and other leagues.
Their “secret sauce,” as the NHL calls it: a database called HANA that computes stats faster than any other before it. It’s the engine that runs NBA.com’s stats pages, allowing fans to dip back into basketball numbers from as far back as 1946.
In September 2014, SAP data scientists in Palo Alto began redesigning the front end of NHL.com, giving the stats pages a facelift and adding new categories while keeping the site’s user-friendly functionality. Photos will run larger. Features are integrated with the stats leaderboards. It all looks a bit prettier.
There have been enhancements to the basic stats that we’re all familiar with – hello, secondary assists, finally – but the most immediate and startling change to the site is the integration of advanced stats, a.k.a. “fancy stats,” which the NHL is calling “enhanced stats.”
For years, fan-run sites had tracked puck possession through stats called Corsi and Fenwick, scraping data from official NHL game sheets to glean information like shot attempts and ice time. Last year, the NHL quietly prohibited data scraping in its official Terms of Service agreement with users, an obvious signal that it planned to start offering these metrics on its own.
The NHL will begin with 30 “enhanced” stats, including five variations of Corsi and Fenwick that account for score situations, relativity and percentages.
(Oh, and they’ll no longer be called “Corsi” and “Fenwick,” either. Say hello to “shot attempts” and “unblocked shot attempts,” as Corsi slips into the same abyss as the Patrick Division and the Campbell Conference. So that’s SAT and USAT; and, going forward, “SAT Even, SAT Ahead, SAT Behind.”)
The site will also offer “extended stats” that include average shot distance; zone starts; goals and assists per 20 minutes and 60 minutes; penalties taken and penalties drawn; and PDO, the advanced stat that’s been used to gauge the sustainability of streaking teams.
The possession stats will initially extend back to the 2010-11 season. That’s when the League updated its real-time stat tracking and had more accurate time on ice numbers.
All these stats will be updated nightly during games, period by period. Eventually they’ll be added to box scores in real time, but that’s down the line.
PHASE TWO (April 2015)
Phase Two of the project is scheduled for an April 2015 rollout, and here’s where we arrive at the aforementioned rabbit hole.
The two primary innovations are stat visualizations and stat search filtering.
At the top of every table of statistics, there will be a line-graph chart that visualizes the numbers for the top three players or teams in that stat category. The X-axis features the dates. The Y-axis features the stats. Think of a stock trade ticker, and you’ll get the idea.
Beside every player’s name will be a check box, easily allowing users to change the players (up to three) being compared in the visualization. You wanna track the goals scored by Alex Ovechkin, Sidney Crosby and Steven Stamkos for the last five years? Go ahead, and see what the trends tell you.
(The ability to embed these charts won’t be available yet, but is in the plans.)
But what if you wanted to know how many goals Ovechkin has scored in the third period to break a tie on the road January on a Wednesday?
The NHL would like to provide you with this information.
The League is scheduled to debut a new search-filtering engine for stats in Phase Two, and it’s going to be epic if you’re someone that enjoys the trivial “split stats” on player pages. (Raises hand…)
For example: Have you ever wondered how many goals were scored in the last three years in the third period at home by a 30-year-old right wing who was born in Canada, played in the Olympics and is under 6-feet tall? Wonder no more – that’s how specific this thing is going to get.
That innovation will give us a good sense of the past; but what about the future?
PREDICTING PLAYOFF SERIES (April 2015)
SAP has a data scientist that created an algorithm for the NHL. It’s one that is going to make TV executives nervous and sports wagers rather happy.
It’s an algorithm that can predict the results of playoff games and series, and it’s coming to NHL.com, scheduled for April.
The algorithm looks at 37 different factors that go into playoff predictions. They including records in teams’ last 10 games, Corsi-for and other advanced stats. It spits out predictions for every playoff game and playoff series, with a breakdown of head-to-head matchups and which team has the advantage in each category.
The scientist tested the algorithm to generate predictions against 15 years of playoff results.
It had an 85-percent success rate.
The site will list around 10 of the 37 metrics that the prediction engine uses, with check marks or stars signifying which team has the advantage.
Just don’t expect to see winning probability percentages. Gotta leave a little something to the imagination.
PHASE THREE (Fall 2015)
By the start of the 2015-16 season this fall, the NHL expects to have further innovations for its stat visualization charts; and, if all goes well with the technology’s rollout, the first introductions of player-tracking stats to NHL.com.
The visualizations will allow for up to three stats to be charted simultaneously for multiple players during a certain time period – a chance to really paint a picture of offensive production, for example, by tracking goals, assists and power play points.
The hope is to hone this for use with fantasy hockey players and leagues.
There are also plans to have the advanced stats integrated into the visualizations by this point as well.
The player-tracking technology could bring a few innovations to NHL.com by next season:
* Real-time live game tracking. The players show up as colorful circles, the puck is a little dot, and you literally watch the players move on the box score on your screen as they move on the ice inside an arena hundreds of miles away – including real-time player-tracking stats.
* The addition of player-tracking stats to stats pages, including average speed of shots, distance traveled and speed of skating.
PHASE FOUR (2016)
NHL COO John Collins has been lauded for being one of the bolder voices inside the league, and he’s trumpeted this idea for years: The digitizing of every NHL box score going back to 1917.
In 2016, the NHL expects the project will be completed, giving NHL.com a completely searchable database of stats going back to nearly a century ago.
Granted, there are some generational challenges in using this data. The 1920s didn’t have assists counted. Shots on goal weren’t tabulated until the 1950s. Plus/minus wasn’t tracked until the 1970s.
Still, the box scores will allow for situational comparisons between players decades apart – how about the number of goals that gave his team the lead in the third period between Gordie Howe and Alex Ovechkin?
But the NHL’s archival information isn’t simply text-based. The League has also been categorizing every goal from every game it can find on film for an incredibly ambitious NHL.com upgrade: The ability to watch every goal a player scores on videos linked from their stats pages.
Currently, every goal of every game is available in box scores. By 2016, the plan is to have every goal from 2007 onward linked from the player’s game logs.
You’ll be able to go to Patrick Kane’s player page and watch every goal he’s scored in the NHL, all in one collection.
Player tracking is the future of NHL stats and, in turn, the future of NHL.com. The single most mind-blowing idea born out of the NHL and SAP partnership, although it could be years away: Real time, 3-D digital recreations of NHL games.
Forget little colored circles representing the players. Think about EA Sports’ NHL series. Now imagine actual games presented in that environment. Rewind plays. Pause them. Parse them. Analyze them. You can see through the eyes of players. You can follow the puck from stick to stick. You can break down every breakdown on a goal – yes, especially you, guy who didn’t back-check.
Combined with real-time stats, it could be the ultimate second-screen experience. Although from the prototype we saw, it might even replace the first one if you don’t feel like paying for GameCenter Live.
Then again, you’ll likely be paying for this too.
The next year will see plenty of trial and error, as the NHL senses what innovations are most important to their fans and which ones need tweaking. (They vetted the first phase with fans at focus groups in San Jose this season.)
Integration with the NHL’s mobile app will be a slow rollout as well.
Some of the stats may be redefined as player tracking tech improves – some in the League still cast a skeptical eye at the stats formerly known as “Corsi” and “Fenwick,” and believe actual puck possession numbers from chips on players will render them extinct. Some of the innovations may face pushback, like the percentage on predictions did – does anyone in the NHLPA want a 3-D environment that can expose their mistakes behind the play?
Then there’s the cost. The first two phases will be free to everyone that visits NHL.com, and said there hasn’t been any discussions about paywalls. But it’s fairly obvious that a large investment from the NHL in all of this data means they believe there’s revenue to be generated from it.
The notion that the more exciting aspects of player tracking and some of the “enhanced” stats would be available on a subscription basis, perhaps packaged with GameCenter Live, isn’t all that outlandish.
The question is whether the NHL’s revolutionary approach to its stats and presentation will be enough to coax fans to ante up.