Advertisement

What We Learned: New NHL stats site has failures at launch

What We Learned: New NHL stats site has failures at launch

There have been a few dozen things written about the NHL's new “enhanced” stats site launching, and one feels safe in assuming the thrust of many of them are, “Uhh, what the hell?”

So: Uhh, what the hell?

The NHL entered the advanced stats era fully and officially on Friday, launching its own engine to give fans more immediate access to the data that has been cultivated for years on the websites of nerds who are truly and thoroughly fans of this sport and wanted to understand the game on a deeper level. Views in that community were a bit mixed; on the one hand, there was trepidation that the whole thing might get screwed up, given how much we know about the NHL's ability to do a Peter Sellers-level job of inadvertently throwing itself down a flight of stairs while trying to show off.

But on the other there was a welcoming feeling because this was validation for the years of hard work in a way. By launching this site, the league was essentially saying, “What you have worked on for years has value;” and indeed, the league did an excellent job of reaching out to many people in the analytics community and working with them to see what they'd like to see on the Official Site.

On a third hand, there was also an understanding that this wasn't really something for the existing advanced stats community. As with anything else, the league — and please recall this is the NHL we're talking about, where goals-against average and wins are the two goalie stats listed on a newly updated “Leaderboard” on the front of the stats page — is catering to the casual crowd of fans, not the deep, obsessive fanatics. Even getting them to give up the ghost on pretending these stats don't exist was a huge step forward.

On a fourth, there's the league's insistence that it's doing something unique, which it absolutely is not. What it feels an awful lot like is a Cortez-type moment. The NHL set foot on land that was already discovered long before, and inhabited by industrious people, and said, “Hmm yes, we own all this.” They built this site, yes, but stood on the shoulders of giants to do it; the league accomplished very little on its own, and it's claiming full credit anyway.

Take for example this interview with Chris Foster, NHL director of digital business development, who said, “There's zone starts, I think those are completely brand-new. And the level of depth that we're doing with primary and secondary assists. I don't think anybody's going to have that much detail.”

Zone starts are of course not “completely brand-new” and Hockey Analysis has been breaking down first-assist leaders in the league for years. As for the later claims of providing penalties-drawn data for the first time ever, well, Extra Skater used to do that before the Maple Leafs bought it, and War On Ice still does. And of course Behind the Net, after which the entire 'advanced' stats era is named, was tracking both long before anyone else

This should have been easy: Take War on Ice or Extra Skater, basically copy the idea behind it with enough changes to avoid a lawsuit, slap an NHL logo at the top. You've just built an excellent website that will help put millions more hockey fans in touch with the concepts underlying success in this sport. Is there a learning curve? Sure. But the NHL just expended a whole bunch of resources building something that people have been making in their spare time — and presumably for much cheaper — to roll out a broken product that doesn't do a good job of what it ought to do. Certainly, not as good a job as the league seems to think.

You wouldn't know it from all the back-patting on Friday afternoon, but the problem is that the league — and one assumes SAP, which worked alongside the league to build the site — seems to have put very little thought into many aspects of the launch.

For example, users can't filter for things like “minutes played” or “games played” at this time, so if they went to see who has the best CF% (sorry: SATF%) in the league when the site launched, they were probably going to be shocked to find that the answer — Bruins' prospect Matt Lindblad, who's gotten all of 16 minutes at even strength this season — was a guy they'd never heard of. All the way down at No. 11 is Pavel Datsyuk, who actually leads the league in this category, and who had played five more games than the 10 guys in front of him combined.

Then there's the fact that you can go and look at a player's SATF per 20 minutes of ES ice time — which, by the way, never states explicitly that it is indeed at ES — and per 60 minutes of ice time, with the latter being the former multiplied by three, and thus not really worth differentiating.

And also, there's the fact that the league leaders in “SAT rel” show how little consideration was put into the ideas behind the stats. The vast majority of the top players in the league are on absolute garbage teams, or have been traded from them, because “SAT rel” is apparently calculated by taking the team's SAT number overall and subtracting that player's on-ice SAT number.  Which is why Buffalo's Nicolas Deslauriers led the league at plus-712; the Sabres may stink, and so does he, but at least he's not on the ice for all the other shot attempts his team gives up.

Let's put it another way: The Sabres are the worst possession team in recent league history, and if you have a stat that puts 15 of their players in the league's top-20 possession leaders, maybe that stat is absolute garbage and doesn't tell users what you intend it to.

Now, to be fair, the league will probably say its stats engine is all being rolled out over time, and that much is true. By April, there will be the ability to visualize stats (which already exists on War On Ice and a number of other sites), as well as the ability to filter the data down to the smallest sample sizes imaginable so that you can draw all sorts of faulty conclusions from the numbers if you finesse them enough to include only Swedish wings who are between the ages of 24 and 32 who shoot left and weigh at least 190 pounds.

At that point the league will also provide data that predicts playoff series, based on a 37-factor algorithm that has an 85 percent success rate. But only in predicting past playoff games, which is a lot easier to do, of course; I can “predict” who will win any individual game or whole series over the last 100 years or so as long as you let me check Wikipedia first. And if someone could develop a formula to predict individual games in any sport with 85 percent accuracy, the sports book industry would have hired them or had them killed years ago.

Fun fact, though: if you want to pick an individual series winner, simply using scored-adjusted fenwick (sorry: score-adjusted USAT, which as far as the NHL is concerned doesn't exist yet) over the last 20 games o the regular season will give you a success rate that pushes 70 percent. Obviously bringing that up to 85 percent to that makes your predictions 21 percent more accurate — if that is indeed what SAP's algorithm can do for future games, which I doubt — but putting money down based on that alone is going to win you a decent bit of cash over the course of the postseason.

Please don't mistake any of this for bitterness. The fact that “corsi” and “fenwick” and PDO are now SAT and USAT and SPSV doesn't really bother me — except to say that it's once again co-opting the work of others and putting a new name on it like Thomas Edison at the height of his Tesla-harassing days; this is a branding issue — and seeing the league embrace the power of these statistics is indeed nice.

And I further understand that I'm in the 1 percent of 1 percent of users who look at this data literally every day of their lives. The average fan isn't me, doesn't have the need for this much information about the sport. This site is likewise not for me; the likelihood that this version was going to supplant War on Ice and Puckalytics and Hockey Analysis and Own the Puck and Puck on Net — y'know, to name a few — for me was basically non-existent. But the fact of the matter is that we've very much gotten off on the wrong foot in terms of helping people understand what these statistics mean and how they should actually be applied, which creates its own problems down the road: their foundation of advanced (sorry: enhanced) stat knowledge being based in bad or faulty information.

There's also the fact that the league published all this data, faulty though it may be, with plenty of caveats that this does not indicate any official position on the efficacy of using possession stats to evaluate teams? “Does it tell you anything? We can't say for sure but if you think it does here it is!” That's classic stuff.

Overall, the league probably would have been better off not rushing to get this done in the middle of the season, I think. Take the extra several months and unveil it next October, make the thing actually usable and intelligible at launch by letting the real nerds get an early look and tell you what you're doing wrong. At that point you can also integrate the player-tracking technology we've heard so much about, and really have something that stands out from the offerings put together and maintained by a small group of people who are essentially hobbyists.

The problem with showing up late to the party and trying to take credit for everyone else's work is that you might not really have the best idea what's going on with it. Certainly this is progress, but it's not the kind of progress for which the league should be holding a press conference.

What We Learned

Anaheim Ducks: No surprise but the Ducks want defense now that they're starting to fall apart a little bit. There's this guy in Toronto I think would be great there.

Arizona Coyotes: Antoine Vermette really isn't doing very much to make people excited about trading for him. Seems like nothing wants to go Arizona's way here.

Boston Bruins: Here's the thing with hyper-reactionary headlines like this: The Bruins were never going to let Malcolm Subban into a position where he even might have become their No. 1 goalie. They have this fellow called Tuukka Rask for the next hundred years, and I guess he's pretty good.

Buffalo Sabres: Tim Murray is apparently turning his attentions to acquiring Ryan O'Reilly, which would theoretically give the Sabres a Nos. 1-3 center depth of Jack Eichel/Connor McDavid, Sam Reinhart, and Ryan O'Reilly for the next decade. Nice way to speed up your rebuild.

Calgary Flames: The Flames won't give a 32-year-old questionably second-line right wing a ton of money this summer? Maybe they are learning.

Carolina Hurricanes: Andrej Sekera is going to cost someone a lot, in terms of both money, picks, and prospects, and there's a pretty good chance he's worth it.

Chicago: Jonathan Toews went in on his teammates after they lost to the Avalanche. After that game, they'd lost four of six, so maybe he's right to do it.

Colorado Avalanche: Speaking of the Avs, this is maybe the nicest goal Brad Stuart has scored in a while:

Columbus Blue Jackets: Sergei Bobrovsky looks close to a return. Just in time for it to not matter at all.

Dallas Stars: Wait, the Red Wings are a “long-time rival” for the Stars? In what possible way is this the case? The Red Wings don't ever think about the Stars.

Detroit Red Wings: Saturday night, the Red Wings entered the third period down two goals, tied the game, went down by two goals again, then won in overtime. First time that's happened since 1934. Doesn't seem like any sort of a fluke at all.

Edmonton Oilers: This is a really great look at how the Oilers develop talent. Or, put another way, how they don't do that. And how things might be changing.

Florida Panthers: If the Panthers had won on Saturday, they would have moved into a playoff spot. That is weird to think about.

Los Angeles Kings: What if this is just how good the Kings are for the rest of the season? Besides the white pants, I mean.

Minnesota Wild: And when you talk about playoff pushes, the one going on in Minnesota right now is fascinating. Devan Dubnyk shouldn't be able to play like this, but man, here we are.

Montreal Canadiens: Seems as though this Pacioretty fellow is a pretty good hockey player. He might score 40 this season. I doubt too many people would have predicted that.

Nashville Predators, America's Favorite Hockey Team: Proof goalie wins are a nonsense stat: Rob Zepp stops 20 of 22 and wins, while Pekka Rinne stops 31 of 33 and loses. Ah well, at least they got the point.

New Jersey Devils: The Devils might try to replace Lou Lamoriello as GM in the next year or two? Wow, only three years too late!

New York Islanders: Wait, are people, like, not-treating the Islanders as one of the best teams in the East?

New York Rangers: Is Mats Zuccarello worth $5 million to you? That's what he's seeking this summer. What a league.

Ottawa Senators: You can go ahead and throw it in, guys.

Philadelphia Flyers: Seems like you can put almost anyone on a line with Claude Giroux and they're gonna score. Maybe we want to dial back the “Jake Voracek is an elite forward” talk as a consequence.

Pittsburgh Penguins: The Pens are looking at bringing back Jordan Staal in the summer. The Staal dynasty is falling apart in Carolina.

San Jose Sharks: The answer to the question posited in the headline is, “Yes,” and it's because of how foolishly they approached things this past summer. They made their bottom-six much worse and moved a dominant power forward to the blue line. Now they're not as good. Hmm.

St. Louis Blues: The Blues' schedule for the month is brutal. Montreal Tuesday, a five-game road swing, three more home games, then six more on the road. We're about to learn a lot about these guys.

Tampa Bay Lightning: Pretty decent little run of success on the road here from the Bolts. Now they're back above .500 away from home (kinda, if you don't call shootout losses actual losses).

Toronto Maple Leafs: If you think Olli Jokinen is going to make a positive impact with the Leafs, of any kind, have some bad news for you.

Vancouver Canucks: Isn't this kind of just what the Canucks do? Beat the Rangers, get creamed by the Devils. Are we really surprised at this point?

Washington Capitals: Picking up a shootout win over the Islanders goes a long way toward making people think the Caps are legit. Not as long as, say, a regulation win, but beggars can't be choosers.

Winnipeg Jets: For real, though, stop doing this.

Play of the Weekend

This is a real nice goal from Niklas Kronwall. Real nice.

Gold Star Award

Feb 21, 2015; Santa Clara, CA, USA; Recording artist John Fogerty performs in the first intermission during the Stadium Series hockey game between the San Jose Sharks and the Los Angeles Kings at Levis Stadium. (Kyle Terada-USA TODAY Sports)
Feb 21, 2015; Santa Clara, CA, USA; Recording artist John Fogerty performs in the first intermission during the Stadium Series hockey game between the San Jose Sharks and the Los Angeles Kings at Levis Stadium. (Kyle Terada-USA TODAY Sports)

John Fogerty destroyed at the Stadium Series game. Best musical guest in outdoor game history. Difficult for anyone to top that.

Minus of the Weekend

Boston Bruins goalie Malcolm Subban is unable to stop the puck on a shot by St. Louis Blues' Petteri Lindbohm (not pictured), of Finland, during the second period of an NHL hockey game Friday, Feb. 20, 2015, in St. Louis. (AP Photo/Scott Kane)
Boston Bruins goalie Malcolm Subban is unable to stop the puck on a shot by St. Louis Blues' Petteri Lindbohm (not pictured), of Finland, during the second period of an NHL hockey game Friday, Feb. 20, 2015, in St. Louis. (AP Photo/Scott Kane)

Malcolm Subban's NHL debut could have gone better. Pulled after allowing goals on three straight shots. Woof.

Perfect HFBoards Trade Proposal of the Week
User “Pen13” isn't too far off.

Clarkson for Semin?

Oh my god if you thought the Toronto media hated Phil Kessel...

Signoff
I was in the Vietnam War. The Nam.

Ryan Lambert is a Puck Daddy columnist. His email is here and his Twitter is here