To carry or to dump, that is the question... and the answers may surprise you
[Last week we pointed you to a post over at NHL Numbers regarding the importance of zone entries and the cutting-edge way they're being quantified and analyzed. What follows here is a guest post by Eric T., the author of that post and the paper that will be presented at the MIT Sloan Sports Analytics Conference, regarding some of the Capitals-specific data that was left on the cutting room floor.]
Introduction: What's this zone entry stuff all about?
It's always bugged me that the existing metrics really aren't very good at assessing defensive skill.
You can look at how many shots or goals the other team gets when a certain player is on the ice, but a low number could be achieved either by playing good defense and breaking up the opponent's plays or by being really good with the puck and prolonging your own possessions to give the defense fewer chances. A good offense might at times be the best defense, but it'd still be more satisfying if we had stats that could tell who does well when the other team has the puck.
So with that aim, last year we started the zone entry project at Broad Street Hockey. The thinking was that by separating the game into offensive zone possessions, we could count how many shots or goals the opponent gets per possession, a number that should be completely isolated from confounding effects of offensive skill.
As we worked up the data, we ran into a surprising result. We had separated out entries with possession (where you carry the puck in or pass it to a teammate) from entries without possession (where you dump or chip the puck deep and chase after it -- not including plays where you just dump and go for a line change). The entries with possession generated more than twice as many shots and goals as the dump-and-chase plays, but once you control for that, it is awfully hard to find players who are particularly good at getting or giving up shots on each entry. Zac Rinaldo might not carry the puck in as often as Claude Giroux, but the team got just as many shots from each of his carry-ins as it did from Giroux's (though Giroux's shots are probably more dangerous).
So with that discovery, we came to the tentative conclusion that most of a team's shot differential is determined by neutral zone play, by getting more offensive zone possessions than the opponent and especially by making those entries with possession of the puck. But we wanted more than just one team's data if we were going to draw a conclusion that strong, so we broadened out the project and looked at other teams.
One of the teams we looked at was the Capitals. So with that lengthy preamble out of the way, let's move on to the reason I'm posting here -- to share what we observed from the Caps.
Individual puck-handling stats
Players can drive shot differential by helping their team win more than their share of the neutral zone battles, so that their team gets more than half of the entries. They can also drive shot differential by challenging well at the blue line, forcing the opponent to dump the puck in. But for this article, I'm going to focus mostly on the individual puck-handling stats -- how often a given player sent the puck into the offensive zone, how they did it, and what the results were.
In the table below, here's what each line means (all stats are 5-on-5 only, and all except the last exclude dump-and-change plays):
- Involvement: When the Capitals got an entry with him on the ice, what percentage of the time was he the one who sent the puck in?
- Entries with possession: When he sent the puck in, what fraction of the time did he carry it in or pass it to a teammate, rather than dump it in?
- Shots/carry: What was the average number of shots (including shots that miss the net but not blocked shots) per play when he carried the puck in?
- Failure rate: What fraction of his attempts to carry the puck in were broken up (turnover at blue line, offsides, puck never gets more than a few feet into the zone, etc)?
- Change rate: What fraction of the time that he sent the puck in did his line immediately go for a line change rather than try to recover the puck?
|Player||Involvement||Entries with possession||Shots/carry||Failure rate||Change rate|
Some miscellaneous observations:
The shots per carry in column doesn't really seem to match any obvious skill tendencies, which supports the observation made on other teams that it was basically random (i.e. the data from odd-numbered games for the Flyers or Wild was useless for predicting how they did in even-numbered games).
The Capitals were not particularly efficient in how they distributed the puck. Over this period, Perreault was generally on a line with Hendricks, and despite being more than twice as likely to gain the zone with possession, Perreault was actually less likely to be the one bringing the puck in. Chimera was the very worst forward on the team at gaining the zone and yet had the puck on his stick as much as anyone. The result is that the team likely did not get the most out of their talent.
Each line ended up with similar rates of having their entry attempts broken up. This suggests that the lower lines aren't really playing things safer; they end up with fewer dump-ins because they aren't as capable of beating the defense, not because they're staying farther away from potential risks. (Although Knuble did have a spectacularly bad year in this regard.)
The lesser lines do appear to be a bit more likely to dump the puck and go for a change. The top four scoring forwards on this list had change rates of 11%, 11%, 9%, and 5%; the other nine were all at least 13% and averaged over 16%.
Impact of coaching
We have about 20 games each under Bruce Boudreau and Dale Hunter. The difference between the two is very clear (stats below are 5-on-5, excluding dump-and-change plays):
|How many of the zone entries went the Caps' way?||50.3%||49.0%|
|How many of the Caps' entries were with possession?||48.8%||47.4%|
|How often did the Caps' attempt to carry the puck fail?||13.5%||18.4%|
|How many of the opponents' entries were with possession?||48.7%||53.0%|
Hunter did not have much impact on the top forwards, but he seemed to do everything he could to keep the bottom six from ever entering the offensive zone in possession of the puck (differences of more than 3% are highlighted):
The difference is awfully clear, and the numbers do not reflect well on Hunter.