Arctic Ice Hockey: FanPosts

Power play shooting percentage

2013-05-18T19:52:43Z

Much research has been done to show how even strength shooting percentage is ridiculously random, on both the individual and team level. As a result, shooting percentage cannot be used to predict future success, whether in the same season or in different seasons.

So I tried to do the same test for shooting percentage on the power play.

Here are the results for the past 6 years (sorry for the weird table, I'm still having trouble getting tables into Fanposts

Team Sh% Sh% Sh% Sh% Sh% Sh%
2013 2011-2012 2010-2011 2009-2010 2008-2009 2007-2008
Anaheim 14.21 10.99 13.68 11.78 13.6 11.27
Boston 10.13 12.82 9.63 11.61 15.07 13.59
Buffalo 9.84 10.76 12.44 12.14 14.19 11.79
Calgary 15.03 12.73 13.28 11.27 11.52 11.55
Carolina 9.48 11.96 11.75 12.63 12.55 14.06
Chicago 11.05 10.66 16.29 12.95 13.62 12.03
Colorado 11.18 11.78 11.71 12.37 11.78 11.54
Columbus 10.11 8.77 8.01 11.52 9.72 11.65
Dallas 12.82 12.69 11.61 12.04 11.11 14.52
Detroit 13.25 11.37 13.67 11.23 13.97 11.57
Edmonton 16.86 16.07 12.06 13.35 13.39 15.11
Florida 14.05 12.8 7.53 11.23 10.39 11.11
LA 16.2 11.55 10.17 14.32 12.62 11.28
Minnesota 12.56 10.31 13.73 14.1 14.14 13.11
Montreal 13.71 9.31 12.82 14.4 12.61 16.52
Nashville 12.57 15.92 10.61 10.39 10.62 11.41
New Jersey 10.75 12.42 8.01 13.47 9.92 9.67
NY Islanders14.43 14.33 10.88 9.84 12.5 10.5
NY Rangers 11.56 12.34 11.67 12.47 9.16 10.28
Ottawa 10.33 13.68 12.38 11.7 14.15 14.55
Philadelphia14.64 13.17 11.44 13.36 17.99 15.19
Phoenix 10.33 11.11 11.11 11.23 11.32 13.62
Pittsburgh 14.98 12.02 11.03 11.14 12.71 13.5
San Jose 11.02 11.61 11.15 13 15.1 11.5
St. Louis 13.78 12.1 13.14 11.95 13.93 13.06
Tampa Bay 13.87 12.5 12.32 11.55 12.69 13.16
Toronto 14.86 13.87 11.8 8.64 11.78 12.42
Vancouver 12.18 12.63 16.46 13.95 11.21 11.6
Washington 20.5 12.05 9.52 16.23 16.25 14.25
Winnipeg 10.59 14 12.64 11.75 14.66 11.61

Here are the r^2 values between each season

2007-2008 to 2008-2009: .20931

2008-2009 to 2009-2010: .10631

2009-2010 to 2010-2011: .00753

2010-2011 to 2011-2012: .00651

2011-2012 to 2012-2013: .05983

Pretty clear that power play shooting percentage cannot predict future power play shooting percentage across seasons.

So why does this happen?

Well first thought may be that roster turnover affects this data. There may be some truth to that, but I would argue that the players that play on the power play (the core of any team) are pretty constant from year to year. You don't see teams letting go or trading power play forwards or defenseman as much as third line and fourth line players.

What I believe is that shooting percentage is just very random from year to year, similar to even strength shooting percentage.

Regardless, power play shooting percentage clearly is not repeatable and seems to be far more influenced by luck than anything else. As a result, teams should not focus on power play shooting percentage when looking at their success on the power play and instead should focus on other metrics than can be proved to be repeatable.

Corsi per goal, the cousin of shooting percentage

2013-05-12T00:50:41Z

I've never seen this before, so I wanted to share this. I was thinking about shooting percentage, and realized that it completely ignores when a player takes a shot and it misses the net or it gets blocked.

Thus, I propose that Corsi per goal (at both the team and individual level) is far more usefull, since it penalizes teams and/or players that miss the net or get their shots blocked at a high rate.

I'm planning on using this stat and seeing if it reveals anything interesting.

Thus, I propose that Corsi per goal (at both the team and individual level) is far more usefull, since it penalizes teams and/or players that miss the net or get their shots blocked at a high rate.

I'm planning on using this stat and seeing if it reveals anything interesting.

Shot blocking percentages for/against and their correlation to winning

2013-05-10T16:41:29Z

Most people in the advanced hockey metrics community are aware that Blocked shots correlate slightly negatively to winning (in other words, teams with a lot of blocked shots will usually have less wins and a worse goal differential than teams with smaller blocked shot totals). This is because a blocked shot for team A actually shows us that Team A didn't have the puck.

However, I thought it would be interesting to calculate the ratio of blocked shots per total shot attempts. I wanted to do this to determine which teams blocked more shots per shot attempt, which teams got more of their shots blocked per shot attempt, and whether that correlated at all to winning and or scoring goals.

Calculating the percentage of shots a team blocks (compared to their total shot attempts total) or the percentage shots a team took that got blocked is very easy. NOTE: all numbers are courtesy of stats.hockeyanalysis.com and are at even strength (5v5) and not including any situations with the goaltender pulled.

Blocked Shots percentage For = Blocked shots by team/ Corsi against *100

Blocked Shots percentage against= Blocked shots by opposing teams / Corsi For*100

(Corsi is equal to shots on goal + miss shots on goal +blocked shots).

First, I did Blocked Shots percentage against for the past three seasons. Remember, this stat shows the rate at which a team's shots get blocked by the opposing team.

The R^2 value for this data was .00085. In other words, no correlation at all.

Second, I will do Blocked Shots Percentage For (past three seasons).

Remember, this stat shows the rate at which a team blocks their opponents shots.

The R^2 value for this data was .01695. Again, too low to show any correlation between shot blocking rates and goal differential percentage.

Conclusions:

From the data, it seems to me that shot blocking rates have little effect on goal differential. While past research has determined that the #of shot blocks is usually lower for a better team, the rate at which an individual team either blocks shots or gets their shots blocked compared to total shot attempts against/for seems to be very close to irrelevant.

I would have put up the graphs with the results but I couldn't figure out how to import an image from my computer.

Any Thoughts?