Zone entry basics
Those of you who have been regular readers for a while know that zone entries are a pet project here. Geoff Detweiler first tracked them for us in the 2011 playoffs, and then in 2011-12 we were able to compare zone entry data across multiple teams.
Advancing the puck forward more often than the opponents do (so they get more entries) and keeping control of it as they enter the zone (rather than dumping it in) are major drivers of outshooting the opponent. Play in the attack ends appeared to be much less of a factor in shot differential; the best lines might get higher quality shots from each possession, but they didn't seem to take appreciably more of them.
This is an exciting transition for advanced stats: we're moving from describing what happened with a specific player on the ice to what the individual might have done to drive those results. The strongest individual contribution to shot differential appears to be what percentage of the time a player's zone entries are by carrying the puck in rather than dumping it in.
The 2013 Flyers
A few of us tag-teamed things to get through the 2013 Flyers zone entry data. Kurt and I each tracked about 10 games, and Jessica Schmidt (@2_for_slashing) did the rest. Let's take a look at what we found for the individual carry-in numbers and how they compared to last year (all stats are 5v5 only):
2013 carry% and rank
||2011-12 carry% and rank
Well, that paints a pretty bleak picture.
For starters, this chart shows exactly how the team's personnel changes at forward hurt their shot differential. They let go of their #1 and #5 drivers of puck possession last off-season, and the players who replaced their ice time came in at #8, #10, and #12.
On top of that decrease in skill, the team as a whole took a step back -- eight of the ten forwards who got significant minutes both years saw their carry-in percentage drop, some of them by double digits. To me, this strongly points to a team-wide issue impacting all of the forwards. Since the coach didn't change, the most likely culprit in my mind would be the defensemen who the forwards are playing with.
In 2011-12, the Flyers had a lot more puck-moving skill on the blue line than in 2013. Matt Carle was allowed to leave as a free agent, and Andrej Meszaros went from playing more than 3/4 of the season to less than 1/4. It's not hard to imagine how that might have led to less clean play coming out of their own end and put the forwards in positions where it would be tougher to gain possession of the opposing blue line.
Moreover, in 2013 the Flyers had a lot of injuries on defense, leading to Bruno Gervais being fourth on the team in ice time. Guys like Kurtis Foster, Oliver Lauridsen, Kent Huskins, Brandon Manning, Andreas Lilja, and Matt Konan all made appearances. Some of those players once were or may some day become good defensemen, but at this point in time, none of them lent the kind of security that might give a forward the confidence to take a risk at the blue line.
And so for a variety of reasons, the Flyers as a whole went from carrying the puck in on 53% of their entries in 2011-12 to carrying it in on 41.5% in 2013.
Is that a big drop? Well, I have 30+ games of data for 14 teams (six from 2011-12 and eight from 2013). The 2011-12 Flyers' 53% is the best of those 14 teams. The 2013 Flyers' 41.5% is the worst of the 14, by almost four points. So yeah, that's a big difference.
And it's made even worse because they went from getting more than half of the entries (50.9%) in '11-12 to getting fewer than half (49.3%) in '13. Those two factors -- both describing neutral zone play -- are why the Flyers went from being 9th in shot differential in '11-12 to 19th in '13.
There is some good news in the data, however. In last year's season review, I said this of Sean Couturier's #11 rank:
Couturier notably did not handle the puck like a playmaker. [...] I don't know whether this is something that will improve sharply with time or a reflection of inherent talent limitations, but he needs to get better at it if he is going to become a top offensive threat.
So I am very pleased to see that we now have some evidence that it's something that improves with age/experience -- while ten players saw their numbers drop, the two who improved were Couturier and Brayden Schenn.
Couturier still needs to get better at it if he's going to have the kind of upper-end offensive contributions we're hoping for, but I'm now cautiously optimistic that this improvement will be a natural part of his development.
In addition, returning to the blue line issues, I have to imagine that adding Mark Streit will help, both because he improves the quality of the defense in general and because he appeared to drive a lot of carry-ins for the Islanders.
It's entirely within reason to guess that Couturier and BSchenn will continue to improve, that adding Lecavalier and Streit will reverse some of the personnel losses from last year, and that the Flyers will see a significant rebound from last year's struggles.
Statguy postscript: How important is the neutral zone?
One thing I wasn't 100% confident in after last year's data was whether the offensive and defensive zone results were meaningful.
It was clear that neutral zone performance (getting more entries than your opponent and carrying the puck in on a lot of those entries) was a big driver of shot differential, but could it really be true that the offensive and defensive zone results really had as little impact as they appeared? It seemed like a pretty shocking result that Rinaldo and Giroux were about equally likely to turn offensive zone possession into a shot, and I want overwhelming evidence before I'll buy into something that seems so counter-intuitive.
One of the tools for looking at this is called a split-half correlation. In that method, you look at whether knowing how a player did in the odd-numbered games would let you predict how they did in the even-numbered games; if it wouldn't, then whatever you're looking at probably isn't really something that players have the talent to control.
What we found last year was that the Flyers' neutral zone scores had a strong split-half correlation, but that it was much weaker in the offensive and defensive zones -- a player got a lot of shots per possession in the odd-numbered games didn't necessarily get more than average in the even-numbered games. So shots per possession didn't look like something they control, at least in that sample.
Still, with this being such an unlikely conclusion, I was holding it up as a question to investigate rather than a proven truth at that point. Then we got a full season of Wild data and again saw nothing particularly compelling -- the offensive zone score reliability was very low, and the defensive zone score, while meeting statistical significance tests, was almost completely at odds with our sense of how good the various defenders on the Wild were.
This year, garik16 took a look at the split-half reliability scores for a few teams. He found that the neutral zone score was repeatable in almost every sample, but that the results on offensive and defensive zone scores remain mixed.
Offensive zone scores looked unrepeatable in last year's Flyers and Wild data, as well as this year's Hurricanes, but looked repeatable in both this year's Islanders and Oilers data. I'd like to add two things to this evaluation:
1) The Flyers offensive zone score repeatability (for players with at least 250 5v5 minutes) was 0.02.
2) The rankings for the Islanders (McDonald, Aucoin, Reasoner, Bailey, Cizikas, Martin, Tavares, Moulson, Grabner, Okposo, Nielsen, Boyes) and Oilers (Horcoff, RNH, Hall, Jones, Petrell, Eberle, Smith, Belanger, Brown, Yakupov, Paajarvi, Gagner, Hemsky) don't match particularly well with my subjective perception of how much offensive talent the players possess. Moreover, they don't match at all with how many points the players scored -- the correlation between OZ score and points for the Oilers forwards was -0.11, compared to 0.85 for NZ score, and the correlation between OZ score and points for the Islanders forwards was -0.45, compared to 0.48 for NZ score.
One way of looking at this: if I tried to tell you that Talbot and Fedotenko were better at generating shots than Giroux and Voracek, or that Aucoin and Reasoner were better at it than Tavares, or that Petrell and Jones were better at it than Eberle and Yakupov...you'd expect some pretty strong proof of something that goes strongly against your intuition. A mixed result on the significance of split-half correlations probably isn't enough to convince you, nor should it be. I don't think anything's proven yet because of the mixed results, but I find "single-season OZ score is mostly random noise" to be the simplest explanation of what we're seeing here, and I come to the same conclusions about the DZ scores. (Tyler Dellow's work on the importance of getting/preventing the first shot attempt relative to the importance of getting/preventing multiple shot attempts provides more circumstantial evidence pointing in the same direction.)
In contrast, both of our 82-game samples and a bunch of half-season samples (except for this year's Canes) all show a significant split-half correlation for neutral zone score, and the rankings come closer to matching what we'd expect. So I find a lot more meaning there, personally.