Score effects and you

Cam Charron
December 05 2013 02:48PM

TH2NSTATSGUY is stating, at first, something that is fact. Despite numbers guys' focus on shots, over the course of a game, the team that outshoots the opposition won't win every game. In fact, they should win less than half of the games.

If TH2NSTATSGUY were really interested in manipulating numbers to make his point, he'd have pointed out that between 2008 and 2012, so five complete, 82-game seasons, teams that out-shoot have combined for a 2982-2272-642 record, and teams that have been out-shot have combined for a 2914-2232-750 record.

Look at games in regulation, and that means teams that have been out-shot have won 2272 of games decided in regulation, and lost 2232. That's under 50%. Why, then, would we focus on shot statistics?

Because score effects (read this article at Arctic Ice Hockey). Basic analytical concepts have made their way into hockey pressboxes. Corsi numbers are brought up by reporters, "attempted shots blocked", I've noticed, has become just as prevalent on broadcasts as "shots blocked". The second point is key because the way the phrase is spun, you can turn a blocked shot into a positive for the shooter as well as the player that blocked it.

There have been definitive pieces on score effects written. When a team is behind, especially in the third period, they will shoot more. Those shots aren't necessarily of higher quality. This is a record of Shots For and Shots Against the entire NHL has taken in certain game states, averaged out to per team season, via the indispensable Hockey Analysis:

  Shots For Shots Against Shot Share Shooting %
Down 2 270 216 55.51% 7.6%
Down 1 348 304 53.33% 7.7%
Tied 678 678 50.00% 7.5%
Up 1 304 348 46.67% 8.4%
Up 2 216 270 44.49% 9.2%

"Shots For when Up by 1" should correspond exactly with "Shots Against when Down by 1", which it does on this table. The shooting percentages are interesting. Teams that press a little harder and fire shots from everywhere are probably more prone to giving up shots off of a rush. Keep in mind this is only 5-on-5 so empty net situations won't apply.

What this means though, is that it's imperfect to judge teams by "Shots For" and "Shots Against". Those numbers are not absolute and have to be taken in context. A team that is better at shot differential with the score tied is going to score more goals in the long run. That is not really up for debate.

When you look at the end of the season, teams that generally out-shoot in tied situations have played enough minutes from behind that their shot overall totals recover. Let's take a look at the top five teams in shot differential between 2007 and 2012, and the bottom five teams:

  Shot Differential
2008 Red Wings 890
2010 Blackhawks 740
2009 Red Wings 691
2012 Penguins 532
2009 Sharks 492
  ...
2008 Oilers -417
2009 Panthers -431
2010 Panthers -469
2011 Wild -477
2008 Thrashers -667

The reward for the top out-shooting team? A dominant Stanley Cup victory. For the worst team at out-shooting? Zach Bogosian. For the second best? A Stanley Cup victory. For the second worst? Jonas Brodin.

As long as broadcasters and writers are going to be discussing Corsi up there [/gestures towards press box] I think it's fair that they're taken in context with score effects.

The other thing that really needs to be understood is that a "win" does not necessarily mean a team outplayed the opposition. Wins aren't always predictive of the future, and neither are shot statistics, but if you're attempting to forecast a team's future, you could do a lot worse than by looking at a team's shot differential while the score is tied or close. If you attempt to use shot statistics to explain the recent past, I hope you have a straightjacket handy.

63811cbf517d2d685ea09e103488ea3a
Cam Charron is a BC hockey fan that writes about hockey on many different websites including this one.
Avatar
#1 JDM
December 05 2013, 06:06PM
Trash it!
0
trashes
+1
3
props

Honest question: do you think TH2NSTATSGUY is just trolling at this point, trying to get a rise out of you and your contemporaries? Because that's the only explanation I can come up with for that tweet.

Avatar
#2 jvuc
December 06 2013, 08:38AM
Trash it!
0
trashes
+1
1
props

I don't follow the logic here.

The implication is that we should be put in a straight jacket for using shot statistic to explain the past.

But then when the a historically heavily outshot team loses a game, then people come out of the wood work and say I told you so and regression to the mean. How can they have it both ways?

Avatar
#3 Matt
December 06 2013, 10:50AM
Trash it!
0
trashes
+1
0
props
jvuc wrote:

I don't follow the logic here.

The implication is that we should be put in a straight jacket for using shot statistic to explain the past.

But then when the a historically heavily outshot team loses a game, then people come out of the wood work and say I told you so and regression to the mean. How can they have it both ways?

People who come out after one or two Leafs losses and say SEE TOLD YOU REGRESSION are being idiots, same as people who come out after one or two Leafs wins and say SEE I TOLD YOU ADVANCED STATS ARE BS. You have to look at longer-term trends. Leafs go from top team in the East at the ten game mark to playoff bubble team halfway through the season? Then you can start talking about regression and looking back at how the team was playing and getting outshot night after night. On a game-to-game basis there are just too many other things going on (also, hockey would be really boring if the game always went to the team with the better Fenwick/Corsi Close%).

Avatar
#4 jdt
December 06 2013, 02:49PM
Trash it!
0
trashes
+1
0
props

Still, your data above supports the 'shoot more - score more' theory. The "down 1" teams scored 51.2% of the goals; and the "down 2" teams scored 50.8% of the goals. Possible moral: "Always play like you are behind." (Of course, before going there, it would be best to actually look at how this played out with respect to how this data relates to the quality of the teams.)

Avatar
#5 ikillchicken
December 07 2013, 04:32AM
Trash it!
0
trashes
+1
2
props

In addition to everything you just said, it strikes me that simply looking at games where a team outshoots vs games where they are outshot will heavily diminish the magnitude of impact. After all, in the vast majority of cases, one team does technically out-shoot the other. But obviously in a 31 to 30 shot game the odds of winning will remain more or less 50/50. That one extra shot only skews the outshooter's chances ever so slightly. Thus, by including all those games in your calculation you're dragging your average wins much closer to 50%. I think, if you just looked at games where teams outshoot opponents by a significant margin (say, 5+ shots) then you'd see a bigger difference in wins.

Furthermore, even as is, 52.4% vs 47.8% means teams that outshoot win close to 10% more games than those that are outshot. That's not a negligible difference by any means. Especially in light of the above.

Comments are closed for this article.