Last week, Ryan Stimson dropped a post over at Hockey-Graphs entitled "Tactalytics: Using Data to Inform Tactical Offensive Zone Decisions," which essentially quantifies aspects of a couple of pretty standard offensive schemes and compares the two. The post is thought-provoking for any number of reasons, including a couple that are of particular interest to us. First off, one of the approaches Stimson (who you'll remember him from the terrific guest post he did in this space using the data he accumulated through his project on passing microstats) takes a look at is a "low-to-high" set-up that's very familiar to Caps fans.
Second, Stimson will be on the next episode of Japers' Rink Radio to talk about his post, offensive systems, the Caps and who knows what else, so read up and be prepared for what should be a(nother) great discussion.
By way of background (and there's a lot of it in this post we did back in January of 2015), you'll recall the emphasis that Barry Trotz has put on the Caps' producing offense via a "low-to-high" approach in the offensive zone - cycle pucks below the dots (low) and kick them out to the point (high) for shot attempts. Low-to-high. But as we noted back in that post 18 months ago...
Unfortunately, point shots are generally very low percentage shots. Unless a team gets a deflection or a rebound, point shots are usually going to be routine saves for an NHL goalie.
"Well that's why the forwards crash the net - to screen the goalie and gobble up rebounds."
Of course that's true, but deflections and rebounds are far less frequent than generally assumed in the common discourse. Relying on a high volume of low-percentage shots hoping for the infrequent high-percentage deflection or rebound doesn't seem like a tenable approach. The percentages are not in your favor by running your offense through the points, but it's damn near axiomatic in the annals of NHL coaching.
So what's the alternative? Well, one alternative is to attack from behind the net (a.k.a. Evgeny Kuznetsov's office), and that's where Stimson's data and analysis come into play because he compared these two approaches and, well, we'll let him take it from here:
Let’s look at the data we’ve collected from just under 400 games this season.
What stands out is that of the shots that are on target from shot assist from behind the net, they are converted 12.8% of the time compared to shots on target from the point that are converted at a much lower rate of 4.1%. Now, the logic behind shots from the point is that a good shot will get through and also create rebounds. While it is true that 4.6% of these shots result in rebounds, you are more likely to generate a rebound scoring chances on a shot assisted from behind the net (6.4%).
To quickly recap: you are more likely to score and generate rebounds by playing behind the net rather than going through the point for offense.
Nothing here should surprise you, but it quantifies the differences here in a way that hadn't been done before (publicly, at least). Not only are you more likely to score and generate rebounds by playing behind the net rather than going through the point for offense, but if we consider goals and rebounds to be positive outcomes (and we do), the behind-the-net approach resulted in a positive outcome 19.2 percent of the time, while the low-to-high scheme was just 8.7 percent successful. That's a pretty sizable difference.
Of course, that difference may be mitigated by how much easier point shots are to come by than shots generated from behind-the-net passes - if it's three times as easy to get off a point shot, then suddenly that 12.8-to-4.1 percent differential in shooting percentage changes the strategic calculus some.
Regardless, it's clear once again that a low-to-high scheme isn't going to generate offense (goals, rebounds, what have you) at as high a clip as a behind-the-net approach. And how has that impacted the Capitals? It's hard to say - there are a lot of moving parts here. But if the Caps think that they're generating rebound chances and scoring dirty goals with a low-to-high offense... they're probably wrong. Here (data via Corsica Hockey), is a look at rebound chances (as both a rate stat and percentage of all shot attempts) at five-on-five for road games (to remove scorer bias) during the 2015-16 season:
Not great (though there aren't rebounds on shots that go in, and the Caps had plenty of those). The Caps generated rebounds at a lower rate than any team in the League outside of Vancouver and Colorado, and had a lower percentage of shot attempts resulting that generated rebounds than any team other than the Canucks. Clearly this is because of an over-reliance on a low-to-high approach. Quod erat demonstrandum.
Well, not so fast (and sorry for mentioning Erat there). It turns out the Caps had among the highest percentage of total five-on-five shot attempts come from their forwards in the League (i.e. they're not just bombing away from the point):
Thanks to Muneeb on that data. Not much "there" there, is there? (Oh, and in case you're wondering, even if you swap out Alex Ovechkin for a mere mortal shot generator, the Caps are in the middle of the pack in percentage).
What we see in the data, then, is that the Caps aren't generating a lot of rebounds at five-on-five, which presumably is something that a team that relies on low-to-high schemes to score would want to be doing (because those are higher percentage chances, though rebound chances, generally, are probably a lot less frequent than people think). But we also see that the Caps may not be as reliant on scoring that way as we (or they) might have thought, which is good, because there are better ways to generate offense... and shots on goal that don't yield rebounds or saves at all.