(NB: I was going to post this much earlier, but even a few weeks late, I think there's enough here worth sharing.)
There is some emerging statistical work being on zone entries (e.g. the MIT Sloan paper here). One possible relationship being examined is between type of zone entry and Corsi+ events. I think it's interesting stuff, but I wondered if counting zone entries alone might tell us something about the game in question.
I wanted to do something simple, yet with some substance, and I wanted to know what it was like to code this live. I thought a scrimmage at the Capitals development camp would be a good opportunity, so I went to the July 11 scrimmage with my legal pad and counted zone entries. (Nope, that's not weird.)
A few non-statistical points:
1) I think the notion of development camp as being more about orientation than instruction is a valid one. (Luckily, the Caps make this open to the public, as this isn't always the case around the league.)
2) Coding live is hard. I can see why reviewing video (really, just a pause button) makes this much easier. We shouldn't forget this. Crap coding = crap data = crap analysis. (Or, just crap brain.)
3) Coding and watching players simultaneously to get any kind of feel as to whose play was standing out was really difficult. By the end, I was just focused on the blue lines and that's it.
So, here's the setup. We start from the notion that puck possession matters a lot to winning hockey games. We also think that the type of zone entry raises the likelihood a positive outcome for one of our current metrics that serves as a proxy for puck possession, Corsi. So, what if we just looked at zone entries between two teams in a particular game? Would we see differentials that correspond to the game's outcome? Can tracking this data tell us anything meaningful about the two teams playing the game?
The procedure was pretty simple. I kept track of period, time remaining, type of entry (carry-in, dump, dump-and-change, chip-in, tip-in, pass-in, shot-on-goal--and by the way, I really only used the last 4 when it was really clear), and 5v5 or 5v4. I then just typed in the data into a spreadsheet and did some simple counting and percentage calculation.
Now, some context. The game was won by the White team 4-0 over the Red team. I thought that (therefore) the White team would have more carry-in zone entries. This sort of seemed right, though my eyes were tracking zone entries not general run of play. I think that focusing on carry-ins might be useful by itself.
Basic aggregate numbers. Not much difference in total numbers and 5v5 play (the actual difference was two zone entries), but for the presentation of results, I'll just stick to 5v5 play. During this time (almost the entire game), there were 144 zone entries, 69-75 favoring the White team. This difference doesn't really tell us anything, but I mention it to give context.
Some specific points:
1) In general, both teams carried in a lot. Out of the 144 total zone entries in 5v5 play, 82 were carries, or 57%. If a player thinks he can carry-in, he will! The reason why dump-and-chase is around is the trap--forward(s) collapsing on the puck carrier toward defensemen sitting at the blueline. Ok, I lied. Actually, there is another reason why dump-and-chase exists: 2-0 leads in 3rd periods (see below).
2) In 5v5 play, the White team (eventual 4-0 winners) had a greater number of carries than dumps. White carries were 59% (44 out of 75 entries). Red carries were 55% (38 out of 69). In Close play (first 98 entries), the numbers were 60% (29 of 48) White carries to 50% (25 of 50) Red carries.
So, this is some (albeit pretty weak) evidence that carry-in zone entries correspond to winning. Ish. (And yes, correct, we could confirm with some actual Corsi numbers, but I didn't count these here).
3) Probably more interesting is to look at UP2+ play (zone entries #99-#146). The Red team carried in 71% of its zone entries, only dumping 3 out of 21 zone entries. The White team still carried in 56% (15 of 27) and dumped 26% (7 of 27). This suggests that the Red team tried to carry-in more, perhaps assisted by a White team that decided to let the Red team gain the zone, and not allow any odd-man rushes.
|First 98 (CLOSE)||RED||%||WHITE||%||TOTAL|
|After 98 (UP2+)||RED||%||WHITE||%||TOTAL|
Data file (.xls) here.
I want to end with two points. First, if development camp is more about orientation than instruction, then the (maybe crappy) data suggests that players (and ostensibly all their coaches) already know that carry-ins lead to better offensive outcomes. The situation determines the amount of risk that is acceptable. Desire to lower risk (the most common strategy for denying odd-man rushes being giving up territory) might correspond to a higher opposition carry-in% (see also the discussion in the MIT Sloan paper). It is true that players and coaches might lean toward the conservative move. Second, in a 4-0 game, it's instructive to note that zone entry differentials were not different in the aggregate. This suggests that the "talent" levels between the White and Red teams were not so great as the 4-0 score indicates.
Since this was a one-off, all of the points here carry the usual caveats. In terms of what this means for the Caps, it would be interesting to see if the raw numbers are at all similar to regular season figures (I know at least a Caps half-season was tracked in the MIT Sloan paper, but the figures are not in the pdf). FWIW, I think that the sibling of the zone exit--puck retrieval after a dump--might be the thing to watch next. If teams are dumping more than they should, those guys who can get the puck after a dump (fast, fearless, though not necessary super skilled) would be real valuable.