Wednesday, May 16, 2007

A look at the numbers

What team leads the postseason in hits? Or how about shot blocking? Faceoffs?

I wanted to know, and figured maybe you did, too, and so I posted the results of a little digging up on Globe on Hockey tonight.


At 10:24 p.m., May 16, 2007, Anonymous David Johnson said...

Most of those stats (particularly hits, giveaways and takeaways) are a joke as there is little or no consistency between game monitors at different arenas. Some of them are so out of whack that they are pretty much useless in raw form.

At 10:32 p.m., May 16, 2007, Blogger James Mirtle said...

As opposed to cooked form?

I'm well aware of the limitations of the statistics, but with things like blocked shots being cited so often, I think there's some useful data there.

At 10:46 p.m., May 16, 2007, Anonymous David Johnson said...

Well, not cooked form but bias reduced form if you attempt to determine amount of bias at each arena and then try to factor it out and you might get more reasonable results.

Blocked shots I haven't looked at for bias but I suspect they are fairly accurate as it is easier to determine if the shot was blocked or not. Either it is, or it isn't. A hit is more open for interpretation and thus much less reliable of a statistic. Was it a hit, just a bump or incidental contact?

At 10:48 p.m., May 16, 2007, Blogger The Forechecker said...

I agree, blocked shots are more likely to be consistently applied, but the hits, takeaway, and giveaways data is so wildly reported that I've ceased working with them. I wonder who we'd contact at the NHL to ask about official scoring standards and what can be done to improve the situation...

At 11:22 p.m., May 16, 2007, Anonymous David Johnson said...

As far as I know the NHL has contracted STATS Inc. to collect those stats so you could try complaining to both the NHL and STATS Inc. but I suspect you won't get anywhere. If anything the NHL is more likely to drop collecting those stats than put the effort and money into training the game monitors to make them more accurate.

At 1:01 a.m., May 17, 2007, Blogger J. Michael Neal said...

THe NHL is unlikely to spend time training people. Stats, Inc, on the other hand, probably does. While they aren't my favorite data collection service (Baseball Info Solutions is very hard to top), they generally do decent work.

The thing I find most interesting about these stats, assuming you give them any credibility, is just how poorly correlated they all are with playoff success. This gets obscured since they are listed in descending order by total, which means that teams that have played the most games are going to be at the top simply because they've had more games to accumulate totals. Rank them on a per game basis, though, and they don't seem to provide much information about why teams are successful.

Caveat: Even the per game numbers are skewed. A rate stat really needs to be normalized to 60 minutes of play rather than 1 game, since a few long OT games will have the same effect as playing more games. That's not a complaint; I am very familiar with what a pain in the ass it is to try to hand collate a bunch of raw data. Just a comment.

At 1:03 a.m., May 17, 2007, Blogger James Mirtle said...

No, you're right JMN, and that's a fair point. Does anyone know an easy way to get the total ice time played for all teams?


Post a Comment

<< Home

Links to this post:

Create a Link


Free Page Rank Checker
eXTReMe Tracker