New Stat Category ("Errors" for players)

[jboknows]

Hi all,

I'll start this with a few points: 1) I'm by no means an expert on advanced statistics and 2) I think +/-, while conceptually making sense, is a pretty weak statistic.

I would love to get peoples' thoughts on a new stat category (perhaps this is covered with some advanced stats # that I don't know how to interpret). I've often wondered why there isn't an "errors" category in hockey. Although subjective, I think it would certainly do a better job than +/- in determining whether or not a player plays a complete game, or if they're on the ice just to pad offensive stats.

The proposal is simple: If a player has a giveaway that leads to a goal, that would be considered an error. When playing 5-on-5, if a player is out of position and it costs a goal, that would also be an error. This is very similar to baseball, when someone makes a brutal throw, drops a simple catch, etc.

It seems pretty unfair for a player who's good defensively to be tagged with a "-" when they do everything perfectly on a shift, but someone else screws up. An error would only be attributable to the players who blew the play. Simiilarly, if no one blows the play and the goalie lets in a goal, there's no impact on anyone (perhaps goalies should get errors for weak goals allowed, although that's even more subjective).

Is this a category that would make sense? Do you see problems with the idea of it? Would it be helpful in assessing a player's ability ?

Click to read more...

 

[Ron*]

Hi all,

I'll start this with a few points: 1) I'm by no means an expert on advanced statistics and 2) I think +/-, while conceptually making sense, is a pretty weak statistic.

I would love to get peoples' thoughts on a new stat category (perhaps this is covered with some advanced stats # that I don't know how to interpret). I've often wondered why there isn't an "errors" category in hockey. Although subjective, I think it would certainly do a better job than +/- in determining whether or not a player plays a complete game, or if they're on the ice just to pad offensive stats.

The proposal is simple: If a player has a giveaway that leads to a goal, that would be considered an error. When playing 5-on-5, if a player is out of position and it costs a goal, that would also be an error. This is very similar to baseball, when someone makes a brutal throw, drops a simple catch, etc.

It seems pretty unfair for a player who's good defensively to be tagged with a "-" when they do everything perfectly on a shift, but someone else screws up. An error would only be attributable to the players who blew the play. Simiilarly, if no one blows the play and the goalie lets in a goal, there's no impact on anyone (perhaps goalies should get errors for weak goals allowed, although that's even more subjective).

Is this a category that would make sense? Do you see problems with the idea of it? Would it be helpful in assessing a player's ability ?

I definitely see problems with at least trying to determine who is out of position. That is so subjective, especially when you factor in rotational defenses and shift changes.

 

[jboknows]

I definitely see problems with at least trying to determine who is out of position. That is so subjective, especially when you factor in rotational defenses and shift changes.

I certainly agree. I think sometimes you can't peg who is at fault, but many times when I'm watching hockey, there is often one player that you can specifically point and say what the hell were you doing. I.e. guys with poorly timed changes, guess puck watching and letting their assignment slip into the slot, etc.

 

[GordieHoweHatTrick]

I'd like to see real-time stats broken down into zones (offensive-neutral-defensive)

 

[Hivemind]

Far too subjective, even for the joke that is the NHL's "real time stats."

 

[hatterson]

If you could consistently apply it across all teams I think it would have some value, although I'd like to expand it beyond just goal events.

One of the reasons +/- is weak is because of sample size. There's just so few goals that occur you can get massive swings based on good or bad luck.

However, I'm not sure exactly how to expand something like that. In baseball an error can only occur when the other team gains a material advantage (ie, one or more additional bases, or an extended at bat) and there's not as many obvious equivalents outside of goals in hockey.

Perhaps scoring chances would work (although that adds another level of subjectivity to it). Maybe include defensive and offensive errors. Defensive errors occur when you make a play that gives the other team a scoring chance, offensive errors are when you cost your team a scoring chance.

 

[Ho Borvat]

I've often wondered why there isn't an "errors" category in hockey. Although subjective, I think it would certainly do a better job than +/- in determining whether or not a player plays a complete game, or if they're on the ice just to pad offensive stats.

The proposal is simple: If a player has a giveaway that leads to a goal, that would be considered an error. When playing 5-on-5, if a player is out of position and it costs a goal, that would also be an error. This is very similar to baseball, when someone makes a brutal throw, drops a simple catch, etc.

The problem I see with assigning "errors" to players is, its pretty black/white in baseball. Shortstop fields a ground ball, and throws it 10 feet over the 1st baseman/outfielder drops a routing fly ball = easy to assign an error to.

I can't really think of any plays in hockey (aside from those weak goals from center ice we see 2-3 times/season) where we can legitimately attribute an error to a player.

Also, going by goals is a faulty system because a goalie like Lundqvist is much more likely to bail out his teammates than someone like Pavelec. And if a d-man makes an error (bobbles the puck at the blueline) its a lot more likely to result in a goal than if a forward makes a mistake (blows a tire behind the other teams net).

 

[jboknows]

Some good responses- thanks for the opinions.

Overall, seems like a failed stat category because:

a) too subjective
b) great goaltending would bail out a player that otherwise should get an "error" - although some had a good suggestion that the error could be based on something other than goals like scoring chances given up (in the defensive zone)

Some people also seem to suggest that the category would not be comparable/fair for defenseman vs. forwards. I should have mentioned this in the earlier post, I was thinking that the stat would be most effective when comparing players of the same position across the league. I.e. looking at Dion Phaneuf's errors vs. Rick Nash would make no sense, but would be more comparable when looking at Staal, etc.

 

[jboknows]

If you could consistently apply it across all teams I think it would have some value, although I'd like to expand it beyond just goal events.

One of the reasons +/- is weak is because of sample size. There's just so few goals that occur you can get massive swings based on good or bad luck.

However, I'm not sure exactly how to expand something like that. In baseball an error can only occur when the other team gains a material advantage (ie, one or more additional bases, or an extended at bat) and there's not as many obvious equivalents outside of goals in hockey.

Perhaps scoring chances would work (although that adds another level of subjectivity to it). Maybe include defensive and offensive errors. Defensive errors occur when you make a play that gives the other team a scoring chance, offensive errors are when you cost your team a scoring chance.

I particularly liked this idea BTW

 

[Tam O Shanter]

I definitely wouldn't want goals against to be a requirement. An error is an error (depending on where you subjectively draw that line) regardless of the scoresheet that 5 teammates have input into.

I actually like this idea, although I would restrict it to VERY OBVIOUS mistakes, and I'd want to keep it to defensive zone and possibly 'last man back' errors. I like when guys give the puck away in the offensive zones, it often means they were driving the net and trying to beat a defender rather than playing the perimeter. It rarely leads to a direct scoring chance. But a man overplaying it at the point, or a giveaway in the defensive end - ya, that's a Brendan Smith - oops, an 'error'!

 

[tarheelhockey]

What you're describing is a lot like the "minus" category of the Subjective +/- project on the Canes board.

http://hfboards.mandatory.com/showthread.php?t=1511657

Surprisingly, if you watch the replay back a few times it's usually not that hard to clearly identify which defenders screwed up. Or maybe the Canes D is just that bad...

 

[NHLPaul]

I think the idea is good and I have thought about something like this as well but I just can't see how it could be accumulated bc of how subjective it is. And yea it can't just be on goals. I think it would have to be on every play of the game

 

[jwhouk]

Giveaways, missed shots, faceoff losses, shots blocked, saved shots (shots on goal less goals scored)... You have a pretty good realm of data as it stands.

 

[Hivemind]

Giveaways, missed shots, faceoff losses, shots blocked, saved shots (shots on goal less goals scored)... You have a pretty good realm of data as it stands.

"Good data" is tenuous.

 

[BroadwayJay*]

Its like taking the worst stat in baseball and applying it to hockey.

 

[charlie1]

What you're describing is a lot like the "minus" category of the Subjective +/- project on the Canes board.

http://hfboards.mandatory.com/showthread.php?t=1511657

Surprisingly, if you watch the replay back a few times it's usually not that hard to clearly identify which defenders screwed up. Or maybe the Canes D is just that bad...

This is what came to mind when I read the OP's first post.