In today’s Toronto Star article (http://www.thestar.com/sports/hockey/2014/03/20/alex_ovechkin_a_perfect_example_of_why_the_nhls_shooting_percentage_is_flawed.html), I talked about Sh% and why it doesn’t tell us enough about how good players are at turning scoring opportunities into goals. Specifically, I pointed out that Sh% excludes shots that never make it to the net because they’re either blocked or miss entirely.
At one extreme of the spectrum was Alex Ovechkin, who only hits the net 50.32% of the time when he shoots the puck. At the other extreme was Jonathan Toews, who hits the net 68.28% of the time.
Phil Kessel wasn’t much better than Ovi, at 54.28%.
Now anyone who watches hockey knows that Ovechkin and Kessel tend to be perimeter players; whereas, Toews likes to play in traffic. So I started wondering if there was perhaps some relationship between where these guys were taking their shots from and what happened next.
Sure enough, Toews was a very close shooter. On average, he shoots from 21.8 feet out, well below our 27 player average of 28.9 feet.
Meanwhile, Ovechkin was the furthest shooter among our group, at 35.6 feet. Kessel (31.7 feet) was not far behind.
What I found interesting was the fact that players who tend to be seen as selfish, puck hogs, defensively irresponsible, etc. by “old time hockey” guys (e.g. Kessel, Ovechkin, Seguin) tend to be far shooters, while so-called “two-way players with grit” (e.g. Perry, Crosby, Toews, Parise) tend to be closer shooters.
A lot more study would be required to make this argument more forcefully, but if there’s a way of quantifying what the non-stats skeptics say can only be seen and judged qualitatively, well…chalk one up for the stats guys.
We then started looking at how players performed together, and this was kind of interesting too.
If you look at Sh% alone, Sidney Crosby (13.49%) is only a shade better than Ovechkin’s 13.39%. But because Crosby hits the net 61.39% of the time, his True Sh% (8.28%) is significantly ahead of Ovi’s (6.74%). Crosby’s linemate, Chris Kunitz, hits the net 59.68% of the time, for a True Sh% of 9.01%.
So how are Crosby and Kunitz doing this? Well, for starters, they too are relatively close shooters. Both of them shoot from 26.6 feet out on average.
What’s really interesting is that the two of them have combined for 136 goals over the past 3 years, which dwarfs Ovi’s 116 goals. And they pulled this off by shooting the puck a combined 1,564 times, which is 157 times less than Ovechkin.
Now I know it’s not fair to compare two guys to one, but there’s something here in my view.
A lot of people have blamed Ovechkin’s horrid +/- (and lack of assists) on his linemates. Several weeks ago, in response to one of our pieces, Scott Cullen made this very argument (http://www.tsn.ca/blogs/scott_cullen/?id=442920).
It’s possible that Marcus Johansson may be an ineffective scorer. And it’s also true that frequent Ovi linemate Nicklas Backstrom (shooting from 34.1 feet on average this season) is a playmaker who likes to work the perimeter.
So it may be that Ovi is simply taking lower percentage shots because a mediocre Ovechkin opportunity is better than a quality Johansson one or giving it to Backstrom to shoot from a mile away.
But it’s also possible Ovechkin’s linemates are shooting so poorly because he’s firing lots of shots from very far away and leaving them with little to score with.
Johansson, after all, was a former first rounder who just turned 23 in October and showed moderate promise the past two seasons, when he was paired with Ovechkin only around 30% of the time (as opposed to almost 60% this season).
Compared to Kunitz, who wasn’t even in the NHL at 23, Johansson didn’t look all that bad.
So I don’t think it’s unfair to question whether some of Ovechkin’s 1,721 shots should have been passes instead, and if they were, whether Johansson might look a bit better.
Or is Ovi simply being used in the wrong way?
I actually find this possibility somewhat intriguing. Should a big guy who has shown an affinity for physical play be playing in traffic more often, thereby freeing up his softer linemates to work the perimeter?
I’m not sure the NHL’s leading goal scorer would get all that many more goals playing that way (because really – in today’s NHL, how many more is he going to get?), but I’m pretty sure his linemates would and his team would be better.
Looking at other pairs, this strategy seems to have worked quite well.
For example, Patrick Sharp, who is not a particularly efficient shooter (True Sh% of 5.99%) often plays with Hossa, Kane and Toews, all of whom shoot from closer than his average of 33.8 feet.
Meanwhile Ryan Getzlaf, who has been a very effective shooter this season (Sh% of 16.38% and True Sh% of 9.06%) shoots from 35.8 feet out on average. However, his total attempts this season (320) are a lot less than those of his frequent linemate Corey Perry (417), who is shooting from 25.3 feet this season.
In other words, when the perimeter guy is shooting the puck in Anaheim, he’s still missing the net more than average, but he’s missing nowhere near as often as Ovechkin, and scoring a lot more.
To me, the numbers suggest that when Getzlaf is shooting from far away, he’s probably got a pretty good look. Because by any standard (Sh% or True Sh%) he’s scoring a fair bit. The rest of the time, he’s getting the puck to the guy who is closer to the net.
A couple of weeks ago, Cullen put Ovechkin on his short list for the Hart Trophy. http://www.tsn.ca/nhl/story/?id=445599
Now to be fair, he acknowledged he was making a controversial argument, and he did put him at #3 behind Crosby and Getzlaf. But his argument boiled down to the fact that Ovechkin was on pace to win the goal-scoring race by 15 goals. It struck us as kind of weird that this dominance was owing solely to the fact that Stamkos, who has dominated Ovechkin in total goals over several seasons now, missed more than half a season.
Add on the fact that Ovechkin’s team is in a dogfight for a playoff spot and his linemates have struggled mightily, and I’m not sure how an MVP conversation gets started.
But if Adam Oates starts running Ovi in traffic? I think we’ll be having a very different conversation then.
The full data on our 27 scorers is below.
| Player |
Shot Attempts |
On Net% |
Sh% |
True Sh% |
Shooting Distance |
| Parise |
986 |
68.86% |
10.60% |
7.30% |
24.7 |
| Toews |
744 |
68.28% |
15.55% |
10.62% |
21.8 |
| Crosby |
676 |
61.39% |
13.49% |
8.28% |
26.6 |
| Moulson |
867 |
60.67% |
13.50% |
8.19% |
24.7 |
| Vanek |
887 |
60.32% |
13.08% |
7.89% |
27.0 |
| Marleau |
1085 |
60.28% |
11.47% |
6.91% |
28.5 |
| Kunitz |
888 |
59.68% |
15.09% |
9.01% |
26.6 |
| Pacioretty |
1141 |
59.16% |
11.56% |
6.84% |
32.8 |
| Eberle |
835 |
59.04% |
14.81% |
8.74% |
25.6 |
| Perry |
1092 |
58.70% |
13.73% |
8.06% |
24.6 |
| Tavares |
1087 |
58.51% |
13.05% |
7.64% |
24.5 |
| Couture |
1036 |
57.34% |
11.95% |
6.85% |
30.5 |
| Hossa |
986 |
57.30% |
12.57% |
7.20% |
29.2 |
| Kane |
1085 |
56.77% |
12.18% |
6.91% |
27.3 |
| Pominville |
967 |
56.26% |
12.87% |
7.24% |
35.0 |
| Seguin |
1133 |
55.69% |
12.04% |
6.71% |
32.9 |
| Ryder |
819 |
55.68% |
14.69% |
8.18% |
29.5 |
| Sharp |
1136 |
55.63% |
10.76% |
5.99% |
33.8 |
| St. Louis |
870 |
55.63% |
14.67% |
8.16% |
28.1 |
| Stamkos |
968 |
55.58% |
19.52% |
10.85% |
29.1 |
| Carter |
974 |
55.34% |
13.17% |
7.29% |
29.6 |
| Malkin |
1112 |
54.86% |
12.79% |
7.01% |
28.1 |
| Neal |
1171 |
54.40% |
13.03% |
7.09% |
30.3 |
| Ryan |
902 |
54.32% |
13.27% |
7.21% |
29.6 |
| Kessel |
1332 |
54.28% |
12.59% |
6.83% |
31.7 |
| Pavelski |
1138 |
51.58% |
13.80% |
7.12% |
31.7 |
| Ovechkin |
1721 |
50.32% |
13.39% |
6.74% |
35.6 |
Congratulations, you've discovered Corsi shooting %, and managed to not at all talk about repeatability.
I normally love your stuff, and am thrilled that we are now seeing these analytics for hockey. But we have to be very careful with the data and the extrapolations that we make from it. I found today's article in the Star, as well as your piece above to be very interesting, but seriously flawed due to some very obvious issues. Especially in the Star article, where you stretch the discussion into the impact of missed shots as opposed to rebounds, etc - which is an interesting conjecture - but how can you even go there when much more significant aspects have been completely missed? In short, shooters who are "snipers" tend to take far more shots that are targeted at at areas that have a higher impact of scoring - such as a few inches from the post or cross-bar in the top corner of the net. But inherently, these shots also result in a higher percentage of misses - becuase the target is essentially a circular area, of which almost three quarters is a missed shot. Contrast this with targeting the center of the net, where the probability of a miss is low, but so is the consequential probability of scoring. You talk about the distance of the shot and position of the shooter above, and these are certainly relevant - but I would argue that it isn't so much where the shot comes from that matters, but where the shot is targeted. And that fact negates this entire analysis of shot attempts versus shots on net. It is probably a strong factor in explaining why shooters like Ovechin and Kessel rank relatively lower in their percentages ratios when you consider overall shots - my presumption would be that they more often target their shots in a way that either result in a goal or a miss. Clearly, by your own stats, more of their shots that are on net go in - and a large factor in this is probably that they more selectively target the most vulnerable areas of the net, ie - they "snipe" - but which also has the consequential outcome that they miss the net more often as well. This represents a huge issue with your analysis - all shots are not created equal.