/cdn.vox-cdn.com/photo_images/680742/GYI0060418301.jpg)
The NHL publishes data for what it calls "Real-Time Super Stats." These stats include most of the subjective judgments like blocked shots, giveaways, missed shots, and hits. Some coaches have structured their game plan around dominating in these areas, and there are players that are sought after specifically for their ability to succeed in one of these categories.
The question I'm going to look into today is how these categories contribute to winning hockey games. I'll be using a simple correlation analysis that looks at multiple years of data. Join me after the jump.
I collected data from NHL.com for four of the real-time super stats: blocked shots, hits, giveaways, and takeaways. I then aggregated the team totals over the last five seasons to come up with a large sample. I also aggregated the total number of regular season points earned by each team over the last five years to serve as a proxy for winning. The table below presents those five year totals:
Hits |
BS |
GV |
TK |
Reg. Pts. |
|
ANA |
8207 |
3854 |
3368 |
2591 |
490 |
ATL |
7088 |
5683 |
3870 |
3923 |
422 |
BOS |
7902 |
5022 |
2659 |
2305 |
451 |
BUF |
6703 |
5573 |
4426 |
2285 |
504 |
CAR |
8690 |
5763 |
2903 |
2955 |
469 |
CBJ |
8111 |
4956 |
2157 |
2371 |
398 |
CGY |
8032 |
4607 |
4029 |
3040 |
481 |
CHI |
7262 |
4712 |
2354 |
2438 |
440 |
COL |
6113 |
6129 |
3426 |
3582 |
449 |
DAL |
9425 |
5004 |
4305 |
3388 |
487 |
DET |
7286 |
4142 |
4045 |
2905 |
566 |
EDM |
6590 |
5972 |
5779 |
3080 |
401 |
FLA |
7830 |
5366 |
3189 |
2492 |
426 |
LAK |
8763 |
5235 |
4562 |
2079 |
408 |
MIN |
6302 |
5071 |
3370 |
2473 |
459 |
MTL |
8943 |
6391 |
4521 |
3601 |
468 |
NJD |
7034 |
4879 |
3525 |
2762 |
516 |
NSH |
6940 |
5015 |
3316 |
3251 |
495 |
NYI |
8601 |
6372 |
3998 |
4151 |
389 |
NYR |
9758 |
5752 |
3244 |
3173 |
473 |
OTT |
9272 |
5694 |
4054 |
3257 |
489 |
PHI |
7545 |
6010 |
4088 |
2812 |
439 |
PHX |
8027 |
5302 |
2647 |
2267 |
417 |
PIT |
8350 |
6048 |
3134 |
2173 |
465 |
SJS |
8639 |
4974 |
4206 |
2639 |
544 |
STL |
7607 |
5660 |
3333 |
2382 |
399 |
TBL |
7370 |
5365 |
3459 |
3012 |
402 |
TOR |
8756 |
5730 |
4386 |
2670 |
419 |
VAN |
6610 |
4807 |
3096 |
2665 |
488 |
WSH |
8090 |
5489 |
4537 |
3424 |
463 |
After collecting this data, I ran a correlation test between regular season points and each of the four individual real-time super stats. That gave me four "r" coefficients.
A quick stat refresher: an "r" coefficient measures the relationship between two variables. It can be between +1 and -1, where +1 indicates a perfectly positive relationship, -1 indicates a perfectly negative relationship, and 0 indicates no relationship at all. As a rule of thumb, an r between 0 and 0.09 indicates no correlation at all; an r between 0.1 and 0.3 indicates a small correlation; an r between 0.3 and 0.5 indicates a medium correlation; and an r between 0.5 and 1 indicates a large correlation. The same rules apply for negative numbers (between 0 and -0.09 means no correlation, and so on).
Here are the four "r" coefficients for the RTSS data:
Before I dive in, I want to talk about how rink bias affects these results. John Fischer over at In Lou We Trust did a marvelous job of looking at the RTSS stats at home and on the road for each team, and found that there were significant differences between the two totals. This would imply that the data reported by the NHL is skewed by the home scorer's definition of what constitutes a giveaway, hit, etc. However, I don't think the problem is that serious, as teams do in fact play very differently on the road, which would lead to significantly different totals between the two venues. And while there might still be a bias for some teams once you control for the differences in home and road play, it most likely is not enough to overcome the complete lack of a relationship in most of the variables.
With that out of the way, the first conclusion is pretty simple: hits, giveaways, and takeaways have no relationship to winning. Coaches who lament that their team isn't being physical enough, turning the puck over too often, or not stealing the puck off the opponent's stick are missing the big picture. They might say this because it makes for a good story line in the media, or because it's convenient to be able to sum up your team's problems in one, short sentence. However, if your team is losing, they're doing something wrong (or just getting unlucky), but it has nothing to do with hits, giveaways, or takeaways. These statistics are, at best, a means to an end, and focusing on their improvement as the solution to your problems is analogous to treating the symptoms instead of curing the disease.
The "r" coefficient for blocked shots was unique in that it demonstrated a strong relationship with regard to winning. The negative sign means that the more blocked shots a team was racking up, the less points they were earning in the standings. This makes intuitive sense as well, as a team that is blocking a ton of shots must be playing in their own end quite a bit. The opposing team thus has the time and space to fire a lot of pucks at the net, which means that you're likely losing the puck possession battle and a lot of games in the long run. These stats show that particular trend.
I think the next logical step is to apply these findings to individual players. Players that excel in takeaways (Pavel Datsyuk) or hits (Brooks Orpik) aren't bad investments because they have a lot of other valuable tools in their skill set. However, going after players only because they're really good at blocking shots, laying out hits, or taking the puck away isn't a wise decision. These skills don't correlate with team success in the long run, and since that's the whole point of playing the game, a team's money could be better spent elsewhere.