Is there a 'Bill James' of hockey?

[Egil]

I think you need more data to analyze hockey correctly.

For example to evaluate goaltending you need to quantify "shot quality" in some meaningful way. In doing so, you are inherantly building in some type of bias. If you are aware of the bias (or believe that your "shot quality" ranking scheme is infallable), then you can statistically evaluate goalies.

To evaluate any skater, you need to quantify defensive ability as well as offensive production. A form of "blame/reward" +/- is probably the way to go, but again requires individual scouting and potential biases of the evaluator. To take it a step forward, combine a "blame/reward" system to the "shot quality" ranking, and you might have a chance.

But pts, +/-, GAA, sv%, etc are all flawed metrics that are nearly useless in evaluating players I think.

 

[konago]

technology guru

He's not Bill James type guy,but
Sharks Assistant to the General Manager,Joe Will, is a technology guru.
http://www.baselinemag.com/print_article2/0,1217,a=126976,00.asp

 

[reckoning]

I've heard roger neilson was really into statistical analysis long before anyone else was.


If anyone's interested in reading some stuff, some of the links i vist are :
The Hockey Project , Puckerings and Hockey Analysis.


You can also go to the "library" section of the Oilers board on here and look at some posts by either Igor or Mudcrutch, they used to post here alot more often and had some really interesting things to look it.




However, if you think stats are of the devil and that your pure, unbiased opinion is the be all and end all, then you probably shoudln't bother.

Daryl Shilling`s Hockey Project site features some very interesting analysis and articles about hockey. I found that site browsing once and nearly spent the entire day reading everything on it. I haven`t seen the hockey analysis site yet or the aforementioned posters on the Oilers board, I`ll have to check those out.

I`m surprised that nobody`s mentioned the Hockey Outsider yet http://www.geocities.com/thehockeyoutsider/home.html. He posts on this board from time to time and has brought up some very insightful interpretations of hockey and it`s statistics. There`s also a group on Yahoo called the Hockey Analysis Group which has plenty of hockey-style sabermetrics, though the math can be a little heavy at times.

The first Bill James type of work in hockey was done by Jeff Z. Klein and Karl-Eric Reif in their outstanding 1986 book The Hockey Compendium.

There`s always going to be some people who will complain about statistics and how "they don`t mean anything and it`s all useless information, etc." Most hockey statisticians don`t claim their work is the definitive final word. It`s all about bringing up new ideas and learning more about the game from researching what`s happened. What harm could come from learning something new?

 

[stanley]

I do agree with this. Hockey is much more of a team game than baseball. But with that said, I do think that statistical analysis can provide some benefit.

Marshall gets a +1 and Jagr gets a -1. That is how you factor in their contribution to the outcome of that play. If Jagr can score 80 even strength points this year that should tell you he is a good offensive player. What if I also say that he is involed in 30% of his teams even strength goals and that as a team they score 15% more even strength goals than the average NHL team...

Gosh, I think plus/minus would be the first metric to hit the chopping block.

 

[Randall Graves*]

Why is it some people are close minded about new ideas?

 

[The_Tactitioner]

wow, I thought I knew abit about stats in Hockey

but after looking at some of those sites, I'm not so sure anymore

 

[KH1]

Yeah, lets keep evaluating guys based on batting average and errors.

You can get just as good a picture of a player from his hits, walks, RBI, runs scored and batting average as you can with the OPS and all that stuff. Plus, some of sabremetrics is just plain dumb (runs scored being more important than RBI for example. That's how decent players in great offenses (Tony Womack for example) get huge free agent contracts and bust.

It should be pointed out that, in the future, you're allowed to skip threads that don't interest you. There won't be a test later.

What's the fun in that? Then I don't get to climb on my high horse and rant about Bill James, who is smarter and more successful than I'll ever be, all from the comfort of my own home?

 

[Krusher55]

You can get just as good a picture of a player from his hits, walks, RBI, runs scored and batting average as you can with the OPS and all that stuff.

Sure, but OPS is one stat, not 5 or 7 or 10. One number is much easier to understand than trying to interpret 5+ stats at the same time. OPS is the best single stat for evaluating a players offensive ability.

Plus, some of sabremetrics is just plain dumb (runs scored being more important than RBI for example. That's how decent players in great offenses (Tony Womack for example) get huge free agent contracts and bust.

No. Sabremetrics isn't why Tony Womack gets a huge free agent contract. Bad scouting and GMing is why. Any sabremetrician will know that if a player has one good year out of 10 and significantly better than all other years, that that year is probably his "career" year.

 

[Krusher55]

you can do anything you want with stats and make them out to be a result of any bizarre scenario you could imagine -- doesnt mean they do anything but if patterns are obvious then you are potentially on to something

i.e -- say you want to look at potential conditioning of players

from a wide view, you can statistically look at how players fair in the specific periods of the 2nd game, where the team plays back to back games.

This is not individual player specific but when teams play games on back to back nights and their opponent does not their record in the second game is:

2002-03: 106-177- 47 0.392
2004-05: 119-166- 56 0.431
2005-06: 46-62 0.426 (through last nights games)

So there is definitely a fatigue factor that comes into play.

 

[Bear of Bad News]

That's how decent players in great offenses (Tony Womack for example) get huge free agent contracts and bust.

No. The free market gives Tony Womack big contracts; sabremetrics has absolutely nothing to do with it. A general manager offers a player a large contract because (in the GM's opinion) the benefits outweigh the price.

And if you're saying that a player can hire statisticians to doctor up his numbers to make him look unbelievable, then you probably also know that each team has several of these people doing the same thing to counter them.

Unless what you're saying is that all baseball GMs are stupid, in which case I might play along for a bit.

 

[Injektilo]

Gosh, I think plus/minus would be the first metric to hit the chopping block.

I think plus minus is good for comparing players who are on the same team against one another and the rest of the teams average, as long as you factor in how tough the minutes are. Igor on the Oilers board did it with Oilers players in 03-04, and it was very interesting. I'd love to see him do it again for this years team at some point.

+- is pretty flawed, but if you account for those flaws, it can be intersting. For example, some of the people on the oilers board were down on Pronger's play so far this year, until it was pointed out that after 21 or 22 games, he'd been on the ice for only 4 goals scored against the Oilers at even strength, which is pretty impressive....

 

[KH1]

No. The free market gives Tony Womack big contracts; sabremetrics has absolutely nothing to do with it. A general manager offers a player a large contract because (in the GM's opinion) the benefits outweigh the price.

And if you're saying that a player can hire statisticians to doctor up his numbers to make him look unbelievable, then you probably also know that each team has several of these people doing the same thing to counter them.

Unless what you're saying is that all baseball GMs are stupid, in which case I might play along for a bit.

I wouldn't necessarily say that all GM's are stupid--Brian Cashman is actually quite smart. It's our good buddy George Steinbrenner who overreacts when he hears people talking about how great sabremetrics is and then goes and tries to buy out the market.

It's not even really the stats theory I have a problem with. It's just that his player rankings in his book show that he clearly isn't the great baseball mind that he makes himself out to be--they are just plain stupid at times.

 

[Injektilo]

This is not individual player specific but when teams play games on back to back nights and their opponent does not their record in the second game is:

2002-03: 106-177- 47 0.392
2004-05: 119-166- 56 0.431
2005-06: 46-62 0.426 (through last nights games)

So there is definitely a fatigue factor that comes into play.

I'd be interested to know where you got those stats from

Or is that the AHL or CHL?

 

[Krusher55]

I'd be interested to know where you got those stats from

Or is that the AHL or CHL?

Sorry, it should read 2003-04.

 

[Injektilo]

Sorry, it should read 2003-04.

ah, i misread it, i thought the first one on the list read 03-04... I was gonna say the increase in winning percentage might have alot to do with the increased number of intra-conference and intra-division games, which means less long distance travelling for teams, which would make them less tired and more prepared for even back to back games.


but since it's down from 03-04 to this season..... any idea why it went up so high from 02-03 and is back down a little this season from 03-04? Like I said, I would have thought it's the opposite...

 

[Krusher55]

but since it's down from 03-04 to this season..... any idea why it went up so high from 02-03 and is back down a little this season from 03-04? Like I said, I would have thought it's the opposite...

It's hard to say. I really need to compare it to the teams overall record. It could just be that the 02-03 stats are low because the teams that played back to back games last year were slightly biased towards weaker teams. It would be interesting to go back a few more years to get a better sample size too. It could be that 02-03 is an anomoly and most years the winning record is closer to what we see in 2003-04 and this year.

 

[Bear of Bad News]

It's our good buddy George Steinbrenner who overreacts when he hears people talking about how great sabremetrics is and then goes and tries to buy out the market.

Alright, I'll give you some more rope - name a player that the Yankees have signed to a big-dollar contract that didn't have big "traditional baseball numbers", but that sabremetrics said would be great. Name one.

 

[KH1]

Alright, I'll give you some more rope - name a player that the Yankees have signed to a big-dollar contract that didn't have big "traditional baseball numbers", but that sabremetrics said would be great. Name one.

You mean aside from Tony Womack? How about Kenny Lofton or John Olerud?

 

[trentmccleary]

You mean aside from Tony Womack?

Forgive me, I haven't read James book... only derivatives thereof. But I've always gotten the impression that a "leadoff man" with a horrible OBP wouldn't cut it according to a Sab.
Oh, and that stolen bases were the most overrated stat in baseball. Especially, considering the A's, Jays and Sox hardly ever steal.

Why would a Sab. give big money to Womack? They should hate him.

 

[LA Blue]

I've heard roger neilson was really into statistical analysis long before anyone else was.

Perhaps, but Neilson's novel contribution to the NHL was intense analysis of game tapes. "Captain Video" was his nickname around the league, and today just about every team has an assistant coach who focuses soley on video.

Matt

 

[Jim Carey Price]

Forgive me, I haven't read James book... only derivatives thereof. But I've always gotten the impression that a "leadoff man" with a horrible OBP wouldn't cut it according to a Sab.
Oh, and that stolen bases were the most overrated stat in baseball. Especially, considering the A's, Jays and Sox hardly ever steal.

Why would a Sab. give big money to Womack? They should hate him.

Yeah don't blame Sab's for Cashman's stupidity. Even if Cashman has a guy to plug Womack's numbers into a computer and make a recommendation, the decision ultimately comes down to Cashman.

Its very likely that Cashman's guy said no, but due to the immense media pressure on the Yankees to get a respectable veteran 2B (who helped the Dbacks beat them in the 2001 WS) instead of going with a "unpredictable" rookie or other unknown player Cashman picked Womack and hoped he'd beat the odds (which he didnt)

 

[DownFromNJ]

Marshall gets a +1 and Jagr gets a -1. That is how you factor in their contribution to the outcome of that play. If Jagr can score 80 even strength points this year that should tell you he is a good offensive player. What if I also say that he is involed in 30% of his teams even strength goals and that as a team they score 15% more even strength goals than the average NHL team. Now we are starting to get an idea of how Jagr compares to his teammates, how his team compares to other teams, and thus with a bit more work how Jagr compares to other players on other teams (offensively). Now, what if he is also a -5. That will tell you a small bit about his defensive ability. But what if I also tell you that the average New York Ranger player is a +5. He produces a lot of offense but his "net" offense/defense rating is well below most of his teammates. Now we are really digging down into Jagr's effectiveness as a defensive player. At this point we should be able to combine his offensive ratings with his defensive ratings to get an overrall understanding of how good Jagr is as a complete player.

True, Jagr gets -1 and Marshall gets +1 on this play. But say Jagr is making a fast break, and makes a beautiful pass to Marcel Hossa, who in turn causes a quick turnover to Brian Gionta (while Grant Marshall is twiddling his thumbs on the other side of the ice) , who skates down the ice and scores. Jagr still gets -1, even though he didn't have anything to do with the goal scoring, and Marshall still gets +1, although he didn't contribute much.

Hockey stats just aren't complete enough. Events involve up to 12 players at once. If Carl Crawford hits an easy ground ball to Robinson Cano, who promply lets it go through his legs, Hideki Matsui isn't charged with an error. Baseball statistics involve one person at a time.

 

[Jim Carey Price]

Hockey stats just aren't complete enough. Events involve up to 12 players at once. If Carl Crawford hits an easy ground ball to Robinson Cano, who promply lets it go through his legs, Hideki Matsui isn't charged with an error. Baseball statistics involve one person at a time.

This is the whole reason I wanted to know if there was a "Bill James" of hockey...and thanks to people posting in this thread, I have found stuff like this:
http://www.puckerings.com/research/plusmin.html

Which, while it isnt perfect, it is better.

 

[KH1]

Forgive me, I haven't read James book... only derivatives thereof. But I've always gotten the impression that a "leadoff man" with a horrible OBP wouldn't cut it according to a Sab.
Oh, and that stolen bases were the most overrated stat in baseball. Especially, considering the A's, Jays and Sox hardly ever steal.

Why would a Sab. give big money to Womack? They should hate him.

What would make a sabremetician's eyes light up is Womack's huge number of runs scored in a contract year. I agree with you though, it doesn't make a ton of sense.

And for the record, I'm not saying that George Steinbrenner and Brian Cashman are sabremeticians (they are far from it actually.) I'm just pointing out how they (read: the Boss) jump on trends only after they've been long established.

 

[Krusher55]

True, Jagr gets -1 and Marshall gets +1 on this play. But say Jagr is making a fast break, and makes a beautiful pass to Marcel Hossa, who in turn causes a quick turnover to Brian Gionta (while Grant Marshall is twiddling his thumbs on the other side of the ice) , who skates down the ice and scores. Jagr still gets -1, even though he didn't have anything to do with the goal scoring, and Marshall still gets +1, although he didn't contribute much.

Yes, that is why you can't take a players +/- in isolation. If Hossa is a bad defensive player that is going to affect Jagr. But Jagr isn't likely to play with Hossa all the time and when he is not playing with Hossa, he is going to get some +'s in isolation from Hossa and when Hossa plays with different players he is going to get some -'s in isolation from Jagr. Over the course of a season a teams better defensive player will rise to the top of that teams +/- list and the teams worst defensive player will fall to the bottom. Of course it is not an ideal statistic, but that doesn't mean you can't derive some useful information from it.

Baseball isn't immune from these situations either. If Player A hits a double and then Player B comes in to pinch run for him, and then Player C hits a home run to drive him in, who gets the run scored? Player B does and all he did was jog from the dugout to second and then from second to home.