I've now finished tracking and adding up the puck possession numbers for Toronto's games against the Washington Capitals and the Tampa Bay Lightning this week. Both were games in which the Maple Leafs won by a large margin, and both were games in which the opposing team won the puck possession battle, likely at least in part due to score effects. However, the details in both games are quite different. So I'll break down each game separately here. As with previous games in my series on puck possession, I'm looking both for insights into how the teams played and to see whether or not other statistics like Corsi, Fenwick, and scoring chances line up with the data I've collected.To recap for anyone who hasn't read one of these articles yet, here's what I'm doing - as I'm watching each game, I keep track of each time either team has possession of the puck in the offensive zone. I define the beginning of a possession as the moment when the offensive team first clearly gains possession of the puck in the offensive zone, and I define the end of a possession as either when the puck leaves the zone, or when the defence clearly has control of the puck without being under significant forechecking pressure. I call the result Time On Attack (TOA). I measure this data only for even strength situations.
First up is the game against Washington on Saturday. Here is the TOA data:
Toronto had a pretty big advantage early on in the game, when the score was tied or they had a small lead. Their TOA decreases pretty dramatically in each of the latter two periods, culminating in a 3rd period in which the Leafs had just a minute in the attacking zone, including just 17 seconds in the attacking zone in the final 12 minutes of the game. To give you some idea of how bad that is, Washington had one shift in the offensive zone that lasted 64 seconds. This wasn't your typical "defensive shell", though the Leafs may have been playing back a bit. Rather, it was Washington dominating possession and the Leafs getting, surprisingly, some very good goaltending from Jonas Gustavsson. You'll see why I say this wasn't a defensive shell when I get to discussing last night's game.
Here's the various possession data compared as ratios:
The ratios continue to line up pretty nicely. Everything here is within 1.4%. I think we can call this data essentially the same. No significant discrepancies exist between any of our data types. Let's move on to Tampa Bay, beginning with the TOA data:
In terms of puck possession, the Leafs got even more thoroughly outplayed than they did against the Capitals. However, I think the defensive shell is a more likely explanation for the effect in this game than the one against Washington, and here's why - in the last 11:30 of the game, the Lightning didn't have a single possession in the offensive zone that lasted more than 7 seconds. As I mentioned way back in the first post on this topic, I've noticed that 10 seconds is roughly the amount of time that it takes to solidly establish possession in the zone. The Leafs spent the last half of the 3rd period last night in an almost perfect defensive shell, barely allowing Tampa to gain the zone, and quickly and efficiently getting the puck back into the neutral zone. It was a pretty impressive performance, and not something we've seen from this team often in the past few seasons.
Here's a comparison with the other data, not including scoring chances because those haven't been published yet for this game:
This is the first time that the TOA data has differed significantly from the Corsi and Fenwick data, and I think score effects are a big factor here. The Lightning had a lot of very brief shifts in the offensive zone in the 3rd period, and they didn't get much offense at all out of those shifts. I won't know for sure what effect score effects have until I've gathered data for more games so that I have a bigger sample to work with. I'd say the result here is interesting, nonetheless.
One interesting thing that I've noticed so far is that the team that wins the possession battle has lost all four games that I've tracked. Score effects may be an explanation for that, so once I finish collecting my 10 game sample I'll go back and see if the data looks any different once score effects are taken into account.
[Note: There's no data here for the game on Sunday that we lost to Carolina because I was having difficulties with my Internet connection that made it impossible to track this information.]