# Individual Point Percentage for 2008-2012

Scott Reynolds
October 15 2012 09:06AM

Photo: Michael Miller/Wikimedia/CC BY-SA 3.0

Late last week, I wrote about individual point percentage, and specifically about the individual point percentage of forwards during five-on-five play in the 2011-12 season. As a brief refresher of the concept (for those who don't like clicking through), individual point percentage is a calculation of the number of times an individual player gets a point (either a goal or an assist) relative to the number of total goals scored while he's on the ice. So, for example, if a player is on the ice for fifty goals-for during five-on-five play over the course of the season and he gets a point on forty of them, his individual point percentage would be 80%.

The idea is that this statistic will tell us which players were driving play in the offensive zone. One of the problems is that, because of the small sample size at the level of the individual season (no player was on the ice for more than 87 goals-for), the results are swamped by luck, which is how you end up with Kyle Brodziak and Matt Halischuk finishing second and third respectively. In order to move the conversation forward, I think we need to have a better sense of how players do over several seasons, which should help to deal with the sample size problem, and give us a sense of what a reasonable range of looks like.

As such, I calculated the IPP of all of the forwards in the NHL over the last five seasons. There were 929 different players who played in at least one game (including about forty named Mike), but just 283 who were on the ice for at least 100 goals-for, and that's the group I'll be talking about here. The average individual point percentage in that group is 70.0% and the median is 69.5%, while the standard deviation is 5.5%. Compared to the individual season, the standard deviation is much lower, which is expected. That the median is exactly the same surprised me a little -- I thought it would be higher since we're looking only at players who established themselves over at least a couple of seasons. Before any further comment, let's take a look at the chart  (the raw data comes from Gabriel Desjardins' behindthenet.ca):

There are far fewer surprises on this list than there was in the list for a single season. The eight players at the very top are two standard deviations away from the mean, and are led by Sidney Crosby, the only player with more goals-for while he's on the ice than games played. All eight players are extremely talented, but I don't know that there would be too many people able to guess that group off the top of their head. With Jordan Eberle and Logan Couture, I'm still somewhat skeptical. Both barely made the cut-off, and I could easily see them falling out of this top group going forward.

There are, of course, still some pretty major suprises. Players like Jamal Mayers and Jordin Tootoo simply don't look like they belong. Of course, both players are also consistently played on the lower lines, and it's possible that they're simply better offensive players than their linemates (though even that's tough to believe).

This group has been above-average over the last five years, and it's quite interesting to me because the list includes a couple of players who are generally regarded as clear passengers. Devin Setoguchi checks in at 71.7%, which is well below Joe Thornton, and only slightly lower than Joe Pavelski. I suppose it's possible that Thornton's penchnat for passing helps to keep Setoguchi's number up, but I was expecting him to be down the list. Alex Burrows pops in at 72.3%, which is well below either of the Sedins, but still a very good showing. Milan Lucic shows even better at 75.2%, which is only slightly behind oft-linemates Marc Savard and David Krejci, and is well ahead of Nathan Horton.

One of the things we discussed in the comments of the last post was players on lower lines making their way up this list by virtue of being clearly better than their linemates but still not all that good. The example of the Vancouver Canucks, then, is pretty curious. The majority of their middle six options are in this list: Ryan Kesler (69.2%), Jannik Hansen (68.4%), Mason Raymond (68.2%), Manny Malhotra (66.4%), Chris Higgins (65.6%), Max Lapierre (65.2%), and Mikael Samuelsson (64.2%) is just slightly off in the wrong direction. I don't recall the Canucks having a much more active defense than most, but it's tough to know how else to explain what's going on here.

This last list has quite a few of the more famous "checking" centers in the league. Martin Hanzal, Stephen Weiss and Dave Bolland for sure, but also players like Boyd Gordon, Jarret Stoll, Sammy Pahlsson, and Eric Belanger. This makes a lot of sense. If these players are being told to emphasize the defensive aspects of the game, they're not going to be hanging around too long in the offensive zone to take advantage of turnovers.

As for Jerred Smithson... who knows. It's not a typo.

### Previously by Scott Reynolds

#2 David Johnson
October 15 2012, 09:59AM

So it looks like a normal range for players is between 60 and 80 with only a handful outside those bounds.

Interesting to see Jordin Tootoo so high and Jerred Smithson so low. Smithson is Tootoo's most frequent line mate over the past 5 seasons. I guess that means Jerred Smithson is so inept offensively he makes Jordin Tootoo look good.

#4 David Johnson
October 15 2012, 12:09PM

Kesler's top 2 wingers over the past 5 seasons are Raymond and Burrows but has also a fair bit of time with Samuelsson and Hansen. Let's take a look at those 5 guys.

Burrows 72.3% Kesler 69.2% Hansen 68.4% Raymond 68.2% Samuelsson 64.2%

Four of those guys are within a couple of percentage points of the mean. Burrows is probably the best pure goal scorer of the group which might explain why he is higher than the others. Samuelsson is quite low, but he may be pulled back by his Detroit days, hard to say without seeing seasonal data, but he is also almost certainly less offensively talented than at least Kesler and Burrows and maybe the others too. But generally, what that is telling me is that none of those guys are more important to the 5v5 offense than the others.

Malhotra and Lapierre are more specialists and not known for their offense so not surprising to see them that low. Malhotra also spent some time playing with Rick Nash which may also have driven down his numbers. Chris Higgins just had several bad years where he didn't produce much offense so no surprise he is that low either.

#5 Rhys J
October 15 2012, 12:53PM

Re: the Canucks, perhaps it's because IPP measures production relative to linemates, meaning it can't distinguish between players that are actually good offensive players and players who are just a little better than the dregs they are playing with. For example, Stephen Weiss ranks fairly low, directly behind Boyd Gordon, David Steckel and Chris Thorburn. However, Weiss has more than double the amount of points in a similar number of games.

Going back to the Canucks, if all three players on a line were more or less equally offensively talented, shouldn't you expect IPP to be distributed fairly evenly, with the C a little in front because centres tend to score more points? If that's the case, IPP really doesn't tell you anything. In fact, it probably punishes players for having good linemates.

Say Kesler gets injured and is replaced by Manny Malhotra. You would expect that the IPP of Kesler's linemates (Booth and Raymond, mainly) would rise a little while total production fell off a bit, simply because Malhotra is not as good an offensive player as Ryan Kesler - to put it another way, if a goal is scored by a line of Booth-Malhotra-Raymond, there is a better chance that Malhotra did not get a point than there is a chance of Kesler not getting a point under the same conditions. Really, offensive talent isn't changing, IPP is just getting skewed by circumstance. Similarly, I think we could expect Sidney Crosby's IPP to fall playing with players that are a little better at generating offense than Pascal Dupuis and Chris Kunitz as there is a greater chance of better players generating offense on their own. Once again, a smaller IPP here doesn't mean diminishing offensive talent since the reduction is fueled more by circumstance than anything.

I'm curious to see a correlation between GFon/Game and IPP. I'd guess that there's a positive one, but not a strong one.

#6 David Johnson
October 16 2012, 07:02AM

@Scott Reynolds

"Linemates are clearly going to have an impact on IPP, but even if the value here is knowing which guys are driving offense on their line, I think that's really significant. It's certainly a lot better than not telling us anything."

I am actually starting to think the above could be significant as well. One area I would like to explore more in hockey analytics is team building and I think IPP could help in that. Knowing which players are integral to a lines offense and which ones aren't can be valuable information from both a coaching and a general managing perspective.

Time permitting, I think I might consider looking at adding IPP to my stats site and maybe even integrating it into my WOWY tables (would be interesting to see if Crosby maintains his IPP when with Malkin or if Malkin takes away some of Crosby's puck time, relatively speaking).

One modification I might make is reducing the value of at least the second assist, maybe make it worth half a point, or even eliminate it completely. Thoughts?