Estimating Tempo and Efficiency

It’s often useful to look at what those in the analytics community are doing for other sports to help you spark ideas on how to look at your own sport. A couple good examples that come to mind:

  • The introduction of Player Radar Charts over at Hockey-Graphs.com. I have been following the analytics community for soccer for some time and wanted to eventually try to develop these radars myself ever since Ted Knutson created them at com (As an aside, StatsBomb writers also follow the hockey analytics community since they’ve looked into topics such as PDO and Corsi for soccer).
  • The time Gabe Desjardins sent out a request for readers to manually track passes for a single game between the Red Wings and Blackhawks late in 2010. The results were a snapshot of what we ought to have in the hockey community, since they are so accessible in the soccer community.

What then if we look at hockey like the game of basketball? Every shot is a shot. Every turnover is a turnover. Every basket is a goal. Without a doubt, KemPom.com is a leader in the basketball analytics community, and the bulk of his research comes down to Tempo-Free stats. Instead of saying Virginia must have a crummy offense since they were 227nd out of 346 teams in terms of points per game, we should consider that they play a style that limits possessions. For every 100 possessions they got, they scored 118.5 points, good for 9th in the country. This is how efficient they were with their possessions, which is a much more reliable predictor of future success than simple points per game and also a better assessment of how good their offense played.

What if hockey teams act in a similar fashion? It’s easy to calculate goals per game, but what if some teams that appear to have tepid offenses are just playing a low-possession game? Conversely, what if we praise teams with a high amount of goals per game but play a fast paced aggressive brand of hockey that lets them see more opportunities? The problem then is in figuring out how many possessions occur in a hockey game.

I did related work in my chapter for Hockey Prospectus 2015-16. I looked at all of the event types in a game log and estimated based on the order of events which team had the puck. For instance, faceoff wins, takeaways, and giveaways all tell us which team has the puck following these events. All types of shots and hits tell me who had the puck directly before these events. This time around, instead of trying to see how long each team had the puck for, I wanted to see how many times the puck changed hands during the game.

I looked at last season’s regular season games and estimated how many times a team had the puck vs how many times the other team had the puck. These two numbers will always be off by one or exactly even in a basketball game, but in hockey they don’t need to be since teams can win more faceoffs or end a possession more often (e.g. having more shots on net that end in a goalie freezing it).

Highest Tempo Teams:

Team Year GP Poss For Poss Against Avg Poss Avg Poss/GP
MTL 2015-16 82 6240 6306 6273.0 76.5
TOR 2015-16 82 6282 6230 6256.0 76.3
PHI 2015-16 82 6241 6160 6200.5 75.6
OTT 2015-16 82 6170 6180 6175.0 75.3
BOS 2015-16 82 6162 6161 6161.5 75.1
L.A 2015-16 82 6225 6049 6137.0 74.8
NYI 2015-16 82 6099 6004 6051.5 73.8
EDM 2015-16 82 5998 6079 6038.5 73.6
PIT 2015-16 82 6024 6000 6012.0 73.3
DAL 2015-16 82 6065 5932 5998.5 73.2
ANA 2015-16 82 6027 5952 5989.5 73.0
WPG 2015-16 82 5868 5956 5912.0 72.1
WSH 2015-16 82 5864 5860 5862.0 71.5
NYR 2015-16 82 5828 5888 5858.0 71.4
CHI 2015-16 82 5788 5907 5847.5 71.3
S.J 2015-16 82 5842 5803 5822.5 71.0
FLA 2015-16 82 5786 5836 5811.0 70.9
STL 2015-16 82 5831 5758 5794.5 70.7
CBJ 2015-16 82 5736 5736 5736.0 70.0
CGY 2015-16 82 5723 5749 5736.0 70.0
NSH 2015-16 82 5691 5725 5708.0 69.6
ARI 2015-16 82 5757 5639 5698.0 69.5
T.B 2015-16 82 5690 5697 5693.5 69.4
VAN 2015-16 82 5483 5789 5636.0 68.7
CAR 2015-16 82 5739 5513 5626.0 68.6
BUF 2015-16 82 5522 5576 5549.0 67.7
COL 2015-16 82 5467 5549 5508.0 67.2
MIN 2015-16 82 5520 5398 5459.0 66.6
DET 2015-16 82 5438 5414 5426.0 66.2
N.J 2015-16 82 5251 5511 5381.0 65.6

 

Most Efficient Teams:

Team Year GP Avg Poss/GP GF GA GF/100 Poss GA/100 Poss GD/100 Poss
WSH 2015-16 82 71.5 252 193 4.3 3.3 1.0
PIT 2015-16 82 73.3 245 203 4.1 3.4 0.7
FLA 2015-16 82 70.9 239 203 4.1 3.5 0.7
DAL 2015-16 82 73.2 267 230 4.4 3.9 0.5
CHI 2015-16 82 71.3 235 209 4.1 3.5 0.5
S.J 2015-16 82 71.0 241 210 4.1 3.6 0.5
T.B 2015-16 82 69.4 227 201 4.0 3.5 0.5
ANA 2015-16 82 73.0 218 192 3.6 3.2 0.4
L.A 2015-16 82 74.8 225 195 3.6 3.2 0.4
NYR 2015-16 82 71.4 236 217 4.0 3.7 0.4
STL 2015-16 82 70.7 224 201 3.8 3.5 0.4
NSH 2015-16 82 69.6 228 215 4.0 3.8 0.3
NYI 2015-16 82 73.8 232 216 3.8 3.6 0.2
BOS 2015-16 82 75.1 240 230 3.9 3.7 0.2
MIN 2015-16 82 66.6 216 206 3.9 3.8 0.1
PHI 2015-16 82 75.6 214 218 3.4 3.5 -0.1
OTT 2015-16 82 75.3 236 247 3.8 4.0 -0.2
MTL 2015-16 82 76.5 221 236 3.5 3.7 -0.2
DET 2015-16 82 66.2 211 224 3.9 4.1 -0.3
N.J 2015-16 82 65.6 184 208 3.5 3.8 -0.3
BUF 2015-16 82 67.7 201 222 3.6 4.0 -0.3
WPG 2015-16 82 72.1 215 239 3.7 4.0 -0.3
COL 2015-16 82 67.2 216 240 4.0 4.3 -0.4
CGY 2015-16 82 70.0 231 260 4.0 4.5 -0.5
CBJ 2015-16 82 70.0 219 252 3.8 4.4 -0.6
EDM 2015-16 82 73.6 203 245 3.4 4.0 -0.6
CAR 2015-16 82 68.6 198 226 3.5 4.1 -0.6
VAN 2015-16 82 68.7 191 243 3.5 4.2 -0.7
ARI 2015-16 82 69.5 209 245 3.6 4.3 -0.7
TOR 2015-16 82 76.3 198 246 3.2 3.9 -0.8

 

In the Tempo table, we see Montreal leading the way. They were involved in games where the puck changed hands the most. They accomplished this feat a couple ways. They were 6th in the league in generating Corsi For events and 2nd in the league in Giveaways. I did a little more digging by looking at past years and conclude that this may be a change in playing style under Michel Therrien just for 2015-16, or just a fluky year since they were 12th in the league in 2013-14 and 13th in 2014-15 in terms of Avg Poss/GP.

We can confirm other trends like New Jersey hanging out in last place for the past three seasons. It’s no secret that they like to play “low event hockey” and this research further validates it. New Jersey and Minnesota, also known for their slower brand of hockey, were also in the bottom 3 in terms of Tempo for the past three seasons.

Looking at the Efficiency table, it’s no surprise to see Washington leading the way. They were 2nd in the league in GF/Poss and also 3rd in GA/Poss which means they were extremely efficient when they had the puck and extremely frustrating to try to score against when they didn’t have the puck. These numbers also tell us how a team plays. Pittsburgh and Dallas played a run n’ gun style of hockey since they could afford to. They scored much more often than they got scored on. Other teams like Tampa Bay and Nashville achieved success by slowing the game down while being more careful each time they had the puck.

This is a first pass for me at looking at hockey through the lens of another similar sport in basketball. There are things that sites like KenPom do that make their work more valuable to the game of basketball than this is to hockey. For instance, he adjusts his numbers based on which opponents a team plays since they play a much more lopsided schedule. Additionally, we ought to look at how predictive these numbers are in determining future success. These numbers mean a lot in basketball, but do they hold the same weight when translated to hockey? Lastly, we should look at instances when a high tempo team plays a low tempo team. Which sort of pace will the game take on?

(All stats come from NHL.com and Stats.HockeyAnalysis.com)

Leave a Reply

Your email address will not be published. Required fields are marked *

*