Frequent poster Doogie posted a link to a cool story in the comments to my lengthy list of observations from the first Canucks game. The story, available here, discusses the efforts of Chris Snow’s efforts to dig up some new and relevant statistical data on the Wild. Some interesting bits and comments:
While the league employees were keeping track of things such as hits, time on ice per shift, turnovers and blocked shots, Wild director of hockey operations Chris Snow was tracking some of what he said will eventually be roughly 10 new statistics never previously compiled.
I’d be curious as to what this is. Any guesses? Off the top of my head, I’d guess that they’re looking at who draws penalties, splits of assists (mentioned later in the story)…tough to say other than that. There’s tons of stuff that could be tracked - I’d like to know how many times a game Martin Brodeur plays the puck, myself and how that compares to other goalies.
“You used to always hear from a scout, ‘He’s a good player’ and that was it. But what did that really mean?” Risebrough said. “That didn’t tell you a whole lot, really.”
Amen. There’s a guy over at CalgaryPuck who says that he wrote the coaching manuals for Hockey Canada. He talks a great game “Phaneuf is a generational talent” and all sorts of organizational behaviour gibberish but I’d be interested in knowing how many wins a player adds to a team. I did some back of the envelope math in 2003-04 and concluded that the Oilers should have dealt Staios - I think I figured that the reward in terms of picks would blow away the games that he helps you win. Logically, if you figure that .250 is about as bad as a hockey team can be (a replacement level team), a 100 point team has 59 points above replacement to be divvied up amongst 20 players. When you think about it, it becomes hard to conclude that any players except absolute elite stars are going to be real losses for teams.
“That’s one stat, for example, I’ve been frustrated by,” Risebrough said. “I don’t think that stat has been very accurate. You’ll see some guy credited with a lot of hits, who you know isn’t a very physical player. So maybe the question should be: What kind of a hit was it? Did it have a big impact at a particular time of the game? Those are the kinds of things, in hockey, that are harder to quantify but maybe with more analysis, we can gain something from it.”
I’m a relatively well known stats guy on the internet and I always find it amusing that people will take shots at me and others interested in stats based on this. Or they’ll make arguments about the worth of a player based on his hits or blocked shots or something. Risebrough is kind of getting it in that he’s asking if the hit has value in terms of when it occurs; I think he’d find that by and large, hits are overrated. Some teams hit a lot and win a lot; others don’t. The NHL tracks a ton of shit that nobody has ever shown has any connection to winning hockey games.
At the beginning of the 2002-03 NHL season, the league discontinued keeping expanded statistics. The feeling among general managers was they were losing too much money in salary arbitration cases because of newly created statistics.
To the best of my knowledge, nobody ever tried to put together a solid argument that none of these stats correlated with winning hockey games. Why they didn’t, I’ve got no idea. It may be that it’s irrelevant, in that arbitration is supposed to ensure that similar players are played the same, but it still strikes me as a very valid argument. I’ve learned bits and pieces here and there about what goes in arbitrations and what submissions look like and I’ve not been impressed.
While Snow said he is still in the infancy stage of trying to determine new statistical values, one area he wants to study is the kind of assists players receive. Up to two assists are awarded on a goal, but “first” and “second” assists are not officially tabulated for each player.
The Wild are stealing my material. I think that it’s a great area for study though. There are a ton of things that can be done with it, in terms of studies like Bill James used to do with matched sets of players etc. Personally, my guess is that in splitting up the credit for a goal amongst the players on the ice, the difference in value between a second assist and a guy who just gets a + is negligible. Snow’s asking good questions though.
Snow will keep a list of all first and second assists on the Wild and as many other teams as he can.
This is pretty much the saddest thing I’ve ever read about hockey. I’ve heard a bit about the NHL’s information tracking programs and they’re archaic. Check out this story on MLB.com and it’s obvious how behind the times the NHL is. What’s worse is that they apparently don’t care and don’t understand the potential value of this data.
Anyway, Chris, if you should happen across this, there’s an Excel script in the Summer 2006 archive. You can fiddle around with it and suck this information off of NHL.com. Easy as pie. You can have this information for every NHL team. You shouldn’t need some bozo with a website to tell you this.
Unless of course, he was just bullshitting the reporter, which seems entirely likely. I’m stunned to learn that this is apparently something that he doesn’t have the resources to do for all teams.
Former Toronto and Chicago GM Mike Smith recently started a statistical analysis company that he is trying to sell to NHL teams.
For those who don’t know about Stats Inc., it was a company that Bill James started that grew out of Project Scoresheet. They started tracking and selling this type of information in baseball. I’ve no doubt that there’s a market for it in hockey as well. The key is figuring out what information is relevant. As a bonus, if teams just buy access to this information as opposed to generating it themselves, my recollection of the CBA is that they don’t have to give it to the players. This will happen, whether it’s Smith’s company or someone else who makes the concept work.
“It’s not just strictly a numbers game like baseball,” Andrew Brunette said. “But you can’t just take goals and assists now to know how good a player is to a team. In ‘Moneyball,’ one of the concepts was, ‘What’s undervalued?’
“In our sport, maybe it’s penalty killing or something in the league that nobody’s paying for and you need. It’s an interesting concept, and I think hockey’s going more that way.”
Smart comments from Brunette. The question that needs to be answerwed here is simple. We know that what teams need to buy is goal differential and they need to do it as cheaply as possible. What they need to figure out is what contributes to goal differential in the various aspects of the game, what is overrated in terms of contribution (my guesses include: big hits, guys who look unreal but don’t produce great numbers and fighters) and what’s underrated.
Lowetide mentioned in another post on this topic that these sorts of stories seem to be popping up more and more frequently. It’s good for hockey I think - baseball has certainly benefitted from this sort of examination and I strongly believe that hockey teams are even more ineptly run now than baseball teams were 25 years ago.
In a related story, Bobby Clarke has apparently waived half of his team. Good use of money there. I guess I shouldn’t be too rough on him - how could someone be expected to figure out that a soft Czech who’d made his career as a perpetually underachieving scorer wasn’t a good fit for a checking role?
The West is now 9-1-3 against the East. The East sucks, hard. I’m telling you, of the best 15 or 16 teams in the NHL, 11 of them are Western Conference teams.