On-ice shooting percentage is sustainable…

Prior to the season Gabe Desjardins and I had a conversation over at MC79hockey.com where I predicted several players would combine for a 5v5 on-ice shooting percentage above 10.0% while league average is just shy of 8.0%.  I documented this in a post prior to the season.  In short, I predicted the following:

  • Crosby, Gaborik, Ryan, St. Louis, H. Sedin, Toews, Heatley, Tanguay, Datsyuk, and Nathan Horton will have a combined on-ice shooting percentage above 10.0%
  • Only two of those 10 players will have an on-ice shooting percentage below 9.5%

So, how did my prediction fair?  The following table tells all.

Player GF SF SH%
SIDNEY CROSBY 31 198 15.66%
MARTIN ST._LOUIS 74 601 12.31%
ALEX TANGUAY 43 371 11.59%
MARIAN GABORIK 57 582 9.79%
JONATHAN TOEWS 51 525 9.71%
NATHAN HORTON 34 359 9.47%
HENRIK SEDIN 62 655 9.47%
BOBBY RYAN 52 552 9.42%
PAVEL DATSYUK 50 573 8.73%
DANY HEATLEY 42 611 6.87%
Totals 496 5027 9.87%

Well, technically neither of my predictions came true.  Only 5 players had on-ice shooting percentages above 9.5% and as a group they did not maintain a shooting percentage above 10.0%.  That said, my prediction wasn’t all that far off.  8 of the 10 players had an on-ice shooting percentage above 9.42% and as a group they had an on-ice shooting percentage of 9.87%.  If Crosby was healthy for most of the season or the Minnesota Wild didn’t suck so bad the group would have reached the 10.0% mark.  So, when all is said and done, while technically my predictions didn’t come perfectly true, the intent of the prediction did.  Shooting percentage is a talent, is maintainable, and can be used as a predictor of future performance.

I now have 5 years of on-ice data on stats.hockeyanalysis.com so I thought I would take a look at how sustainable shooting percentage is using that data.  To do this I took all forwards with 350 minutes of 5v5 zone start adjusted ice time in each of the past 5 years and took the first 3 years of the data (2007-08 through 2009-10) to predict the final 2 years of data (2010-11 and 2011-12).  This means we used at least 1050 minutes of data over 3 seasons to predict at least 700 minutes of data over 2 seasons.  The following chart shows the results for on-ice shooting percentage.

Clearly there is some persistence in on-ice shooting percentage.  How does this compare to something like fenwick for rates (using FF20 – Fenwick For per 20 minutes).

Ok, so FF20 seems to be more persistent, but that doesn’t take away from the fact that shooting percentage is persistent and a reasonable predictor of future shooting percentage.  (FYI, the guy out on his own in the upper left is Kyle Wellwood)

The real question is, are either of them any good at predicting future goal scoring rates (GF20 – goals for per 20 minutes) because really, goals are ultimately what matters in hockey.

Ok, so both on-ice shooting percentage and on-ice fenwick for rates are somewhat reasonable predictors of future on-ice goal for rates with a slight advantage to on-ice shooting percentage (sorry, just had to point that out).  This is not inconsistent with what I  found a year ago when I used 4 years of data to calculate 2 year vs 2 year correlations.

Of course, I would never suggest we use shooting percentage as a player evaluation tool, just as I don’t suggest we use fenwick as a player evaluation tool.  Both are sustainable, both can be used as predictors of future success, and both are true player skills, but the best predictor of future goal scoring is past goal scoring, as evidenced by the following chart.

That is pretty clear evidence that goal rates are the best predictor of future goal rates and thus, in my opinion anyway, the best player evaluation tool.  Yes, there are still sample size issues with using goal rates for less than a full seasons worth of data, but for all those players where we have multiple seasons worth of data (or at least one full season with >~750 minutes of ice time) for, using anything other than goals as your player evaluation tool will potentially lead to less reliable and less accurate player evaluations.

As for the defensive side of the game, I have not found a single reasonably good predictor of future goals against rates, regardless of whether I look at corsi, fenwick, goals, shooting percentage or anything else.  This isn’t to suggest that players can’t influence defense, because I believe they can, but rather that there are too many other factors that I haven’t figured out how to isolate and remove from the equation.  Most important is the goalie and I feel the most difficult question to answer in hockey statistics is how to separate the goalie from the defenders. Plus, I believe there are far fewer players that truly focus on defense and thus goals against is largely driven by the opposition.

Note:  I won’t make any promises but my intention is to make this my last post on the subject of sustainability of on-ice shooting percentage and the benefit of using a goal based player analysis over a corsi/fenwick based analysis.  For all those who still fail to realize goals matter more than shots or shot attempts there is nothing more I can say.  All the evidence is above or in numerous other posts here at hockeyanalysis.com.  On-ice shooting percentage is a true player talent that is both sustainable and a viable predictor of future performance at least on par with fenwick rates.  If you choose to ignore reality from this point forward, it is at your own peril.


This article has 9 Comments

  1. Great post David.

    1. I was reading A. Ryder’s NHl 2011 review again last week and he pointed out that….

    “There is pretty strong evidence that shot quality is not very descriptive on offense. To the extent that one can sift through the statistical noise it turns out that the shooter‟s skill supersedes the circumstances of the shot. However on defense, after you have effectively averaged across a lot of shooters, shot quality turns out to be an important descriptor of a team.”

    This seems to recognize the importance of a shooter’s skill ..? I thought of you immediately.
    Also, having witnessed the Canucks firast hand over the last 5 years,
    a time when they have consistently put up high shooting % I don’t need convincing..the Sedins specially have the skill you allude to.

    Also, back to are discussion on shot quality Ryder sure uses it for evaluating defence!

    “Expected Goals is the sum of goal probabilities, across all shots. It is a weighted shot / Corsi / Fenwick assessment, the weights being shot quality (the probability of a goal given the circumstances of the shot). It is such a useful measure that it has been borrowed, repackaged, renamed and sold to one or more NHL teams.”

    3. Why I think Quick is overrated.
    LA was ranked high last year in Ryder’s defensive ranking(#2).
    It is essentially the same team if not stronger defensively without Johnson, as you so correctly pointed out during the trade.

    Ad, seeing how they have played against Van in playoffs, we can safely say Quick in this case, that Quick is facing easier than average shots like he did last year. (I will wait to read Ryder’s 2012
    review to confirm).
    But I do admit that less shots doesn’t necessarily mean easier time for goalie on a general basis.

    Great work..keep it up!

  2. David

    You’ll recall, in the comments at my site, that I made the following comment:

    The comparison of Crosby and Moen isn’t really apt, since we’re talking about lesser players.

    In any event, I grabbed the data off your site so I could check. Of players who’ve been on the ice for 1000 shots or more over the past four years (481 guys), only 46 can we say with 95% certainty that their true talent on-ice S% over the past four years is not between 7% and 8.5%.

    I’m not denying that there’s a talent here – Tom Awad convinced me. The problem that you have is that most players aren’t Moens or Crosbys. They fall somewhere in between.

    The list is in the comments.

    I’m curious – for the guys who don’t make that list, does S% predict goals better than Fenwick? It kind of seems to me that you’re declaring that you’re correct on the basis of a point that was conceded.

    Can you run those charts and exclude the guys for whom we’re confident that there’s some skill/lack of it?

    1. You are right, if we just take the “average” players the correlation falls apart but this makes perfect sense. If we take a bunch of guys of similar on-ice Sh% talent, the difference between them will naturally be shot rates. For example, I took all players who had 2007-10 shooting percentages between 7.0% and 9.0% which is approximately 1.0% on either side of average and shooting percentage and got the following.

      Middle Sh% with Sh% predicting future Sh%: r^2 = 0.1224
      Middle Sh% with Sh% predicting future GF20: r^2 = 0.064
      Middle Sh% with FF20 predicting future GF20: r^2=0.4227
      Middle Sh% with GF20 predicting future GF20: r^2=0.4077

      So yes, if we are comparing similarly talented on-ice shooting percentage players FF20 is a perfectly fine comparison metric. The thing is, the opposite is true too. If we take all the players with FF20 between 12 and 14 we get the following:

      Middle FF20 with FF20 predicting future FF20: r^2 = 0.1264
      Middle FF20 with FF20 predicting future GF20: r^2 = -0.0284
      Middle FF20 with Sh% predicting future GF20: 0.2854
      Middle FF20 with GF20 predicting future GF20: 0.3021

      FF20 is a garbage predictor of future GF20 and even a poor predictor of future FF20.

      My beef is that very rarely do I ever see anyone take into account shooting percentage when evaluating players and whenever I bring up shooting percentage or use goals as a player evaluation metric I usually get some criticism for it. Yes, over half a season shooting percentage is useless, but the majority of players we have far more than half a season of data to work with. Even worse, I see people using shooting percentage or PDO as a proxy for luck which is complete nonsense without taking into consideration what a players on-ice shooting percentage talent is. I am not saying you do it, but a lot of people do.

  3. How are you deciding to use 12 to 14? IIRC, if you cut out the 46 guys, you still have a massive group of players. Your 12-14 FF20 group is going to be way smaller, no?

    1. It was somewhat arbitrary that I selected 12 to 14 but there were actually a lot more 12 to 14 FF20 players than 7-9 SH% players, but as I type this I realize I shouldn’t have used 7-9 for shooting percentage since I used zone start adjusted data where the league-wide shooting percentage is about 9%, not 8%. Let me redo this this way. Since there were 108 players in the 12-14 FF20 group I will take the middle 108 players in the SH% group. Doing so I get:

      Middle Sh% with Sh% predicting future Sh%: r^2 = 0.033
      Middle Sh% with Sh% predicting future GF20: r^2 = 0.0159
      Middle Sh% with FF20 predicting future GF20: r^2=0.1717
      Middle Sh% with GF20 predicting future GF20: r^2=0.1232

      In the above group 2007-10 on-ice shooting percentages range from 8.2% to 10.75%. Of the whole group shooting percentages range from 5.54% to 12.39%. The larger group contains 169 players, the smaller group 108 players so we have cut out 36% of the players, or approximately anyone outside of one standard deviation (which would be 8.05% to 10.82%). One standard deviation from the mean for FF20 ranges from 11.80 to 14.06 so the 12-14 selection is pretty close to the one standard deviation range. Thus the ranges are similar.

      The conclusion is, when you restrict for Sh%, FF20 becomes the better predictor (over Sh%), when you restrict for FF20, Sh% becomes the better predictor (over FF20). Overall the importance of shooting percentage and fenwick are probably pretty close to equivalent which is not inconsistent with Tom Awad’s work in his “What makes good players good” series which may actually give a slight edge to shooting over corsi.

      Being good at fenwick/corsi is still a good thing, it’s just not the only thing.

  4. “very rarely do I ever see anyone take into account shooting percentage when evaluating players”

    I actually think everyone’s on the same page here. For example, a player’s expected shooting percentage is a very true talent – R^2 = 0.82: (that’s May 2010)


    Shooting percentage itself is not as resilient because the small number of goals scored is very luck-driven. Vic Ferrari usually recommends three year averages for individual shooting percentage.

    Where there isn’t much talent is in influencing the shooting percentages of your teammates. (That is, the portion of on-ice shooting percentage that is not due to your own shots.)

    btw, if you use zero-mean variables (and zero intercept) in your regressions, then you’ll be able to see how much a given talent regresses to the mean.

    1. “Where there isn’t much talent is in influencing the shooting percentages of your teammates.”

      I am not sure I agree with that 100% but we don’t need to debate that right now because, to be honest, it is irrelevant. Regardless of how a player can drive on-ice shooting percentage, if he can drive on-ice shooting percentage, we must consider it when evaluating players. Failure to do so will only result in poorer quality player evaluation.

      1. Don’t be silly. If a player’s influence on his on-ice shooting percentage is 100% his own shooting talent, then that means the only thing that influences the on-ice shooting talent of his teammates is their own shooting talent.

        1. Not sure what you are getting at here, but like I said, it is irrelevant whether a player can influence his teammates individual shooting percentages or not (for the sake of argument I am willing to assume he can’t, or whatever you believe). All that matters is that shooting percentages (individual or on-ice) is an extremely important talent when it comes to scoring goals, probably about equal to that of shot generation (as shown above). Thus, you can only get a true understanding of a players true ability and value through a goal based analysis, preferably a multi-year analysis.

Comments are closed.