Worst Stat for Measuring a Player

jc17

Registered User
Jun 14, 2013
11,015
7,733
+/- gets ripped on so much but its not as bad as people make it out to be. I needs context like everything else.

If nothing else, it highlights things that other numbers might not track. I always find its really interesting when looking at stats of individuals on a team most of the team is like -15 and you have 1 or 2 guys that are even or in the small positives. I don't think that should be overlooked.

In other words, most numbers can be useful if they are relative to a players team and teammates.
 

Canadiens1958

Registered User
Nov 30, 2007
20,020
2,773
Lake Memphremagog, QC.
Loses

Loses or team loses. Wins or championships may gloss over individual weakness so you may have a slight over evaluation. Loses too often preclude looking at strengths or correctibles that if corrected or moved to better circumstance produce elite player results.

Prime example Patrick Roy in junior.
 

93LEAFS

Registered User
Nov 7, 2009
33,841
20,898
Toronto
+/- gets ripped on so much but its not as bad as people make it out to be. I needs context like everything else.

If nothing else, it highlights things that other numbers might not track. I always find its really interesting when looking at stats of individuals on a team most of the team is like -15 and you have 1 or 2 guys that are even or in the small positives. I don't think that should be overlooked.

In other words, most numbers can be useful if they are relative to a players team and teammates.
differentials that drastically differ from a teams baseline tend to be caused by zone starts and QOC more than anything else, so i'd say its fairly useless, especially of the box score stats. No stat is truly useless, they just need context and how there being evaluated. For example shooting % if you look at a guy with a high percentage and you think its because he has an elite shot your using it wrong, more likely it will revert to the mean, same logic applies to a low one.
 

Brock Radunske

안양종합운동장 빙상장
Aug 8, 2012
16,787
4,701
The worst stat for measuring a player is Imperial, 100%. Using fractions is less accurate than using the Metric system.
 

Canadian Game

Registered User
Jul 18, 2005
4,937
1,945
Ontario
I always take GAA with a grain of salt. You can be the best goalie in the world but with a crappy defence you're bound to let in more goals than a crappy goalie who's not facing many quality shots because of a great defence.
 

Brock Radunske

안양종합운동장 빙상장
Aug 8, 2012
16,787
4,701
I always take GAA with a grain of salt. You can be the best goalie in the world but with a crappy defence you're bound to let in more goals than a crappy goalie who's not facing many quality shots because of a great defence.

I think GAA and sv% need to be considered together because the same thing could be said about a great sv% but a lot of GAA. Maybe the goalie has a tendency to stand on their head all night but let in 3 softies (like Reimer and Bernier in recent years)
 

Michael Farkas

Grace Personified
Jun 28, 2006
13,350
7,830
NYC
www.HockeyProspect.com
That's correct. It's weird to me that save pct. is put on this pedestal while GAA is automatically dismissed as a "team stat"...it doesn't make a ton of sense to me...at least not to the degree that it is worshiped...
 

SladeWilson23

I keep my promises.
Sponsor
Nov 3, 2014
26,735
3,220
New Jersey
That's correct. It's weird to me that save pct. is put on this pedestal while GAA is automatically dismissed as a "team stat"...it doesn't make a ton of sense to me...at least not to the degree that it is worshiped...

Completely agreed!

SV% is too reliant on the number of shots goalies see in individual games. GAA might not be a good evaluating tool for goalies, but it tells us more than SV%.

Any goalie with a low GAA and low-ish SV% is probably just falling victim to a good team defense in front of him. Brodeur and Quick for example are not really any worse at stopping pucks than Lundqvist or Hasek.

Also people tend to overrate how much of a difference 0.01 in SV% actually is. A .920 goalie isn't necessarily better than a .910 goalie. Stopping 92% of shots instead of stopping 91% of shots can be explained through several different reasons other than a difference in ability. It's 1 goal allowed on every 12.5 shots vs 1 goal allowed on every 11.1 shots.
 

Doctor No

Registered User
Oct 26, 2005
9,249
3,970
hockeygoalies.org
Also people tend to overrate how much of a difference 0.01 in SV% actually is. A .920 goalie isn't necessarily better than a .910 goalie. Stopping 92% of shots instead of stopping 91% of shots can be explained through several different reasons other than a difference in ability. It's 1 goal allowed on every 12.5 shots vs 1 goal allowed on every 11.1 shots.

In your example, the 0.910 goalie allows goals 12.5% more often than the 0.920 goalie. That's a meaningful difference.
 

hatterson

Registered User
Apr 12, 2010
34,950
12,220
North Tonawanda, NY
If you're trying to name a worst stat you probably have to define a couple things. First, what is the list of stats you're choosing from. There's basically no end to the number of literally useless stats you can make, so you have to draw a line somewhere. Attendance on Tuesday night away games? Face offs taken on nights where there's an active space walk from the ISS? Number of days it takes to grow a full beard?

Second, how do you define worst? Is it simply the stat that would result in the least informative rankings when ordered? If so you're basically looking for a stat that has as close to zero correlation as possible with a mythical "player quality" rating. Does worst instead mean most easy to misuse?
 

Doctor No

Registered User
Oct 26, 2005
9,249
3,970
hockeygoalies.org
I know it can make a significant difference, but it doesn't mean there is a gap in puck stopping ability.

True - a simple binomial distribution example shows that two goaltenders of identical ability can have widely different save percentages in a full NHL season. Of course, the same can be said for goal totals, or point totals, or other non-goalie statistics that we use as gospel. Often, the difference between a 40-goal scorer and a 20-goal scorer is opportunity.

On the other hand, retrospective save percentage does show exactly what happened perfectly well. Whether or not there's a gap in puck stopping ability, the 0.910 goaltender *did* allow 12.5% more goals than the 0.920 goaltender.
 

Rebuilt

Registered User
Jun 8, 2014
8,736
15
Tampa
+/- gets ripped on so much but its not as bad as people make it out to be. I needs context like everything else.

If nothing else, it highlights things that other numbers might not track. I always find its really interesting when looking at stats of individuals on a team most of the team is like -15 and you have 1 or 2 guys that are even or in the small positives. I don't think that should be overlooked.

In other words, most numbers can be useful if they are relative to a players team and teammates.

True that. The Edmonton Oilers are a minus -34 goal differential yet Taylor Hall is a plus 2 , Leon Draisaitl is a plus 4 . Even Connor McDavid is only a -2

Brandon Davidson is an eye popping +7. :amazed:

Guys like Eberle, Yak and Nurse are all grovelling around -10-12.
 

SladeWilson23

I keep my promises.
Sponsor
Nov 3, 2014
26,735
3,220
New Jersey
True - a simple binomial distribution example shows that two goaltenders of identical ability can have widely different save percentages in a full NHL season. Of course, the same can be said for goal totals, or point totals, or other non-goalie statistics that we use as gospel. Often, the difference between a 40-goal scorer and a 20-goal scorer is opportunity.

On the other hand, retrospective save percentage does show exactly what happened perfectly well. Whether or not there's a gap in puck stopping ability, the 0.910 goaltender *did* allow 12.5% more goals than the 0.920 goaltender.

I don't disagree. I'm just saying if I'm a GM, and two goalies were there to choose from, I'm not overpaying for the .920 guy if I can have the .910 guy at a bargain.
 

Ad

Upcoming events

Ad

Ad

-->