Interesting Stats: The Shootout is a Crapshoot

Interesting Stats: The Shootout is a Crapshoot

By: HFBoards.com
November 15th, 2007 · Click to Comment

Cliff’s Notes: A team’s performance in the shootout has almost no correlation with that team’s performance in regulation.

All statistics in this post are from previous seasons. There isn’t a large enough sample size to look at 2007-08.

Some teams have performed better in the shootout than others. For example, Tampa was 10-2 last year, while Carolina was 0-5. Interesting, in regulation time, Tampa was actually below 500, with a record of 29-33-20, (meaning that 20 games ended in ties and went to OT) Similarly, teams that were strong in regulation were sometimes weak in the shootout (e.g. Ottawa, Detroit), while sometimes they were strong (e.g. Buffalo).

Anyways, I asked the question - can you predict how well a team will do in the shootout based on how well they do in regulation time? In other words, how is a team’s performance in the shootout correlated to their performance in regulation?

I started by calculating every team’s 2007-08 winning percentage in (a) shootouts, (b) 4-on-4 OT, (c) regulation - where games that end in ties are counted as ties.

Here are the records in the Northeast (REG = regulation)

Ottawa - 0.616 REG, 0.400 OT, 0.250 SO
Buffalo - 0.598 REG, 0.625 OT, 0.714 SO
Toronto - 0.506 REG, 0.500 OT, 0.364 SO
Montreal - 0.500 REG, 0.667 OT, 0.545 SO
Boston - 0.384 REG, 0.667 OT, 0.692 SO

I computed these values for the entire league, and then computed their correlations.

Just a quick word on correlations (skip this if you already understand). Correlations measure how much variables vary with each other. If variable X always goes up when Y goes up, and always down when Y goes down, then the correlation is positive (with a maximum value of 1 if they move in perfect unison). If they move in opposite ways, the correlation is negative (to a minimum value of -1). If they don’t move together at all, then the correlation is 0.

Okay, so how are regulation records, OT records and shootout (SO) records correlated?

Correlations
0.025 - The correlation between regulation and SO performance
0.108 - The correlation between regulation and OT performance
0.230 - The correlation between OT and SO performance

This is interesting. It means all of the following:

(1) A team’s performance in the shootout is barely related to their performance in the actual game
(2) A team’s performance in 4-on-4 OT is better related to their performance in the regulation game, but not too strongly
(3) A team’s performance in the shootout is better related to their performance in the 5-minute OT period than the 60-minute game.

Before I go on, I should say something about the absolute magnitude of these correlations. We can attempt to correlate other hockey variables. We can ask - How does a team’s goal production in one period correlate with their performance in other periods?

Example correlations

0.419 - Correlation between goals scored in 1st period and 2nd period
0.145 - Correlation between goals scored in 1st period and 3rd period
0.517 - Correlation between goals scored in 2nd period and 3rd period

So you can see here a few things:

1) These correlations are fairly large
2) The correlations are stronger between adjacent periods than the 1st vs. 3rd. This is likely because a team that jumps out to leads will have less incentive to score in the 3rd.

Anyways, these example numbers put the REG/OT/SO numbers above into perspective.

If regulation success does not determine shootout success, the interesting question becomes - what does? Interestingly, teams’ success in the shootout can vary wildly from year to year:

0.068 - Correlation between teams’ 06-07 SO record and their 05-06 shootout record

Thus, since teams don’t perform consistently in the shootout it seems there is at least a strong element of luck (of course rosters change from year to year).

The other obvious explanation is goaltending. It seems obvious to me that there isn’t a great discrepancy in shooters from team to team. So if there is a difference in personnel, it is likely to be in nets. This is somewhat hard to measure. There were only six goaltenders who faced 25 shootout shots in both 05-06 and 06-07: Brodeur, Lundqvist, Thomas, Dipietro, Giguere and Luongo.

0.218 - Correlation in their SO save pct’s from 05-06 to 07-08

This seems to indicate that some goalies just might be better at shootouts than others, although the sample size (in terms of # of shots, and # of goalies) is quite small. It would certainly make sense that some goalies would be better at this aspect of the game than others. Interestingly, the order from best to worst was: Lundqvist, Thomas, Dipietro, Brodeur, Luongo, Giguere.

The other explanation for shootout success is that some teams are just more “clutch”. This, to me, is not really an explanation, just a catch-phrase used to describe a phenomenon. I don’t put much stock in to it. There are several shooters on each team, and they don’t seem to perform consistently. If you’re talking about “clutch” goaltending, we may as well just call it goaltending, and i’ve already addressed this factor.

In Short:

my view on the shootout is this: it has absolutely nothing to do with regular hockey, as can be seen in the lack of a correlation between success in the shootout and success in 5-on-5 hockey. Most variance in shootout results probably comes from plain old luck, although goaltending surely plays a role. But it’s not the absolute quality of goaltending - just the specific trait of stopping breakways. Further, I doubt that there is much difference between the quality of shooters on any given team, although it would be difficult to test this due to small shooting sample sizes.

The correlation between OT and SO results is interesting, but it is hard to read too much into. There were only 117 games settled in the OT frame last year. This could indicate the importance of goaltending in both endeavours (although goaltending is also important in regulation of course - perhaps because there are more odd man rushes in OT), or it could simply be a statistical anomaly. I doubt it says anything meaningful about fatigue.

HabsCast Hockey Blog

Hockey News That’s Disturbingly Awesome.