A few weeks ago I came across a HockeyProspectus article by Gabe Desjardins that highlighted the difference in shot percentage between defensemen and forwards and related it to average shot distance. The article concluded that when forwards take the same shots from the distances that defensemen take shots, there is a fairly insignificant increase in shot percentage, thus proving that the difference between a forward's average 9.6 Sh% and a defenseman's 4.1 Sh% is largely inversely correlated to shot distance.
I decided to take a look at the Devils shot splits and distances compared to those of team that showed similar issues (though not as bad) in terms of the ratio of Corsi% to goals scored: the Los Angeles Kings. In addition, I took a look at the Penguins, a team without a terribly high Corsi% that scored many goals, the Blackhawks, a team that was high in both catergories, and just for kicks, the Ottawa Senators who had the lowest Sh% in the league.
The Kings led the league in Corsi% last year, well ahead of the second place Devils, but only ended up slightly ahead of the middle of the pack in goals scored (much further ahead than the Devils at 2nd to last).
I used Desjardin's model numbers to weigh the team's expected shooting percentages in all situations using http://www.extraskater.com ... Here were my results:
|2013 Season||New Jersey||Pittsburgh||Chicago||Los Angeles||Ottawa|
|D SF %||26.36%||22.48%||24.86%||27.38%||32.23%|
|F SF %||73.64%||77.52%||75.14%||72.62%||67.77%|
|D Sh Dst Avg||37.78||40.17||36.23||36.31||37.59|
|F Sh Dst Avg||22.4||23.47||21.99||22.53||30.3|
|Sh Dst Avg||26.45||27.22||25.53||26.30||32.65|
|Split Weighed Exp Sh%||8.15%||8.36%||8.23%||8.09%||7.83%|
This is, of course, presuming that each player on the team is a replacement level shooter. Looking at it from this perspective, coaching strategy and player positioning would probably make the main impact in these numbers.
The first thing that jumps out is that F SF% seems to correlate with Sh%.
The relatively high F SF% of both the Chicago Blackhawks and the Penguins, two teams that had very high team Sh% and G/60 is indicative of this. Although incredibly strong in possession, The Devils and Kings didn't have quite as good splits, and could be expected to suffer in terms of Sh% because of it. Another thing you'll notice is along with Ottawa's very low 66.7 F SF %, their forwards shot from an average distance of 30 ft: much higher than the other teams. This alone could explain Ottawa's absurdly low team shot percentage from last season, but I don't have enough data to create a model for weighing shot distance to Sh%, so i can't account for exactly how much that would effect it. Even presuming that shot % lowers exponentially as opposed to arithmetically with greater distance, we can say it would have a large impact.
The better teams in terms of F SF% on this chart typically did better than the other ones in terms of Sh%, which leads me to believe that the result shows a correlation. I'll be analyzing all 30 teams from last season to see what I can come up with.
Does this explain why our shot percentage was as low is it was, or why Pittsburgh's was so high? No, because plenty of it could be both personnel and luck, and the fact that the percentages don't take into account the powerplay and penalty kill, but it does show that our strategy is probably negatively impacting our shooting.
There are some flaws to my method of calculating expected team shooting percentage such as:
- Gabe's percentages and the percentages he references are from 5v5 play, while my numbers are all-situations, thus my expected shooting percentage fails to account for increase caused by the power play
- There is no model in place to weigh distance deviation from league average team Forward/Defensemen shooting spots for their respective splits
- iCorsi splits and G/iCorsi are not being looked at, which may give a more accurate view of team shooting proficiency
- Shooting talent is not accounted for
As of now, I cannot account for three these of flaws because I cannot obtain forward/defensemen splits on shots for in purely 5v5 situations, I do not have the data that Desjardins does on each individual shot's distance so I can't tell how much a 22 ft shot would effect a percentage as opposed to a 30 ft shot, and because I do not have a set of data accounting for each individual shot attempt (including the distance it was taken at, by whom, and whether it scored or not).
Shooting talent can be weighed (albeit very roughly) using the 14% improvement that Gabe noted forwards had from the same distances as defensemen. I'd suspect that the difference between a Sidney Crosby, and a Dainus Zubrus from 20 ft out wouldn't be all too different, but it'd be nice if I could make sure (I can't). I'd probably have to make a sort of 'grading system' for each individual forward on each individual team based upon career shooting percentage in which I would add 14% to 9.6, leave it as is, or take it away.
Let me know what you think.