1. 4.

Monday, December 7, 2009

Who Is The Most Productive NHL Player? – The Colonel

What is the measure of a Hockey skater’s contribution to his team? Plenty of statistics define goaltenders and their contributions. But what about the players who skate in front of them? While highly knowledgeable Hockey fans will say such enlightened comments as ‘it depends on their position and how they play it,’ many others will tell you it’s how many points they rack up.

At On Goal Analysis, we have a tradition of looking at things with a different twist. While we like to key on points, sometimes they are misleading because theoretically speaking the leading point scorer in the NHL still might play on a team that does not even make the Playoffs. And yet, maybe points themselves just need a slightly different emphasis to make their true significance understood.

That’s why we are recommending for your consideration Points Per Shift – PPS – as a new statistic to use when analyzing who is the most productive player on the ice. PPS analyzes how many Points Per Game (PPG) each player provides divided by the average number of shifts he takes in order to tell you what he brings to the Great Game each and every time he goes over the boards. It also makes Shifts Per Game (SPG) more relevant to the average fan of the game.

Looking at the current Top 30 scorers and from them determining the Top 10 players in terms of their PPS is our aim today. We will also delve into the top Defensemen and Rookies to see how they compare.

Points Versus Points Per Game

After going to NHL.com’s statistical section, it is easy to view all statistics for scoring and provide a list of Top 10 Scorers. As of games ending on Super Saturday night, 5 December when 28 of 30 teams hit the ice, the Top 10 in terms of points were:

Rank Player
1 Joe Thornton
2 Marian Gaborik
3 Sidney Crosby
4 Anze Kopitar
5 Henrik Sedin
6 Corey Perry
7 Patrick Marleau
8 Dany Heatley
9 Nicklas Backstrom
10 Brad Richards

This is where most people would stop in their analysis of who produces the most for their team.

Some, however, prefer to place their money on PPG because not all of the players above have had an equal chance to get on the ice. (From this list you have a high spread of 31 games for Joe Thornton down to 26 games for Marian Gaborik.) So if you were looking at PPG instead, you would have the following Top 10 list of players:

Rank Player
1 Marian Gaborik
2 Alex Ovechkin
3 Ilya Kovalchuk
4 Joe Thornton
5 Evgeni Malkin
6 Sidney Crosby
7 Brad Richards
8 Corey Perry
9 Ryan Getzlaf
10 Henrik Sedin

These stats tell you who is giving the best overall effort per game. Ovechkin, Kovalchuk, Malkin and Getzlaf are not in the Top 10 points getters but provide a Top 10 PPG for their team.

Management, Coaches and Scouts, however, are either in constant search of who can produce the best effort each and every shift, or try to wring it out of them when they play. Their statistic of choice should be Points Per Shift, the PPS.

Points Per Shift – The Top 10

As stated earlier, PPS = PPG / SPG. Of note in conducting this analysis are a few facts. First is that the average SPG for the Top 30 scorers as of 5 December is 24.8767. Getting 24+ SPG in almost every case means a player is on the Power Play, the Penalty Kill or both. Otherwise, 45 second shifts for teams rolling four lines would be 20 SPGs per player in a game.

A second note is that the Top 30 scorers’ average PPS is .04559 for games ending 5 December. You have to use the five digits to separate players’ rank without ties as this is a very finite statistic to decipher. It also became apparent when crunching the statistical analysis the more games and shifts a player has, the lower his PPS tends to go.

Based on the Top 30 players’ stats as of 5 December’s games, here is how the Top 10 PPS rank:

Scoring Rank Player PPS
14 Alex Ovechkin 0.07186
27 Ilya Kovalchuk 0.07065
22 Evgeni Malkin 0.05860
29 Maxim Afinogenov 0.05556
2 Marian Gaborik 0.05468
3 Sidney Crosby 0.05368
9 Nicklas Backstrom 0.05182
10 Brad Richards 0.05148
5 Henrik Sedin 0.04835
1 Joe Thornton 0.04813

There are a few points to discuss from the table above. First and most obvious is the actual players’ scoring rank to their left with 40% of the list NOT among the Top 10 point-getters in the NHL. Also of note are the PPS’ for both Ovechkin and Kovalchuk. Harking back to the post by OGA’s Big Tex entitled “Maurice Richard and the Myth of 50 in 50 (Part 2)” a new stat was offered called Statistical Supremacy. In terms of goals, it was “…scoring a minimum of 50% more goals than the nearest competitor….” I offer PPS displays Statistical Supremacy by netting a PPS 1.5 times or more greater than the data set of players’ average PPS. In the Top 10 above, both Ovechkin and Kovalchuk display Statistical Supremacy as their PPS’ are 1.58 and 1.55 times the Top 30 scoring average respectively.

Looking individually at the Top 10 produces some additionally interesting notes:

1. Alex Ovechkin has a PPS average of .07186 in only 21 injury- and suspension-riddled games this season. Ovie’s games played (21) and SPG (19.9) are both below the average which contributes to his high SPG. Also of note here is that in 21 games, Ovechkin has less than a minute total of PK time which lowers his SPG a bit. If you go strictly on scoring totals he comes in with a Number 14 ranking, but he stands as our most productive player at 5 December’s conclusion based on PPS.

2. Ilya Kovalchuk’s PPS is .07065 stacking him in the mix at Number 2 based on 20 total games with days dressed lost for injury. Kovalchuk does this on an average of only 18.4 SPG, a number along with total games played that rest below the overall Top 30 scorer averages. He also skates less than 10 seconds per game on the PK. Based on our measurement of PPS, Kovy is currently the second most productive player in the NHL despite being only the Number 27 scorer.

3. Evgeni Malkin ranks as the Number 3 most productive player with a PPS of .05860 over the span of an injury-shortened 23 games. His SPG of 21.5 and PPS are better than the norm, and he stands as the highest of the Top 30 scorers to average 20 or more SPGs. Malkin skates in all situations which aides his point totals, but otherwise might normally harm his overall PPS. This fact gives you more insight into his overall team value.

4. The fourth most productive NHL player is Maxim Afinogenov. What? you ask. The same Afinogenov we saw not producing last year? Yes, that Maxim has a PPS average of .05556 in 26 games. He is the only one of the top four players in this Top 10 list who has not lost any games to injury. He nets this PPS while averaging only 18.0 SPG, the least of any of our pool of scorers. Between this player and the top three above, he also holds the least average Time On The Ice (TOI) on the PK (1 second per game) which is likely aide in increasing his PPS average.

5. At Number 5 is the current goal scoring leader, Marian Gaborik. He has given the Rangers a PPS of .05468 over 27 games through 5 December with one game lost due to injury. Gaborik provides this with the second highest SPG average on our Top 10 List at 26.7. This high SPG number at his PPS tells you he provides a huge personal effort for the Rangers on the ice. Were his shifts to be cut back and his performance remain on average at its current pace, he would rank higher on our list.

6. Number 6 in PPS and Number 3 in total points is Sidney Crosby. He holds a PPS of .05368 in 29 games. Crosby does this in an average of 23.1 SPG while playing in all situations. He, too, has lost games for injury this year.

7. Nicklas Backstrom of WSH skates in at Number 7 with a PPS of .05182 PPS in 29 games. He is one of only three in this group whose SPG is higher than the average (25.8), again telling you how much he contributes to his team’s success.

8. Brad Richards comes in at Number 8 with a .05148 PPS in 27 games. Richards plays in all situations, averaging 23.7 SPG, which makes all of his stats better than average. He is also the last of our Top 10 listing to carry a PPS of .05 or greater.

9. Standing as the Number 5 scoring forward in the NHL after 5 December’s games, Henrik Sedin holds down the Number 9 spot in our group with a .04835 PPS in 29 games. Like the majority of our Top 10 PPS’, he rests below the Top 30 Scorer average with only 24.2 SPG.

10. And rounding out our list at Number 10 is Joe Thornton was the leading assist-getter in the NHL after games completed on 5 December. Of note with him is that his PPS of .04813 in 31 games comes in with the highest SPG (26.8).


On Goal Analysis offers the statistic of Points Per Shift (PPS) as a measure of how much a player provides for his team each time he hits the ice. The Top 10 players’ PPS is interesting in that does not follow right down the line with who has simply scored the most points. In fact, the top four on our list rank between Number 14 and Number 29 in overall scoring. That is because in addition to providing true relevance to the Shifts Per Game statistic, it tells us in general that a lower number of SPG for players than the average allows them to provide more effort for their team each time they skate. In particular for this list at this point in the season, it also shows us that both Alex Ovechkin and Ilya Kovalchuk display Statistical Supremacy in this marker over their peers. AS a bottom line, PPS serves to underscore that each of these 10 players is a leader in productivity on their team.

On Tuesday, 8 December, in “PPS Part 2” we will discuss the PPS of Defensemen and Rookies who have played 20 or more games…

Take me to www.ongoalanalysis.com


Nathan said...

I think another point that's important to touch on (and perhaps you will in the next parts of this PPS series) is that PPS is really only useful for assessing what (in most cases) we already know about a player in more detail. I think that PPS would be useful for fine-tuning, which I imagine is why you say it's a good metric for a coach.

There are a few things that I think should be pointed out about PPS:
- PPS is "rewarded" by being off the ice!
- PPS could be perceived as a positive stat when looking at strict scoring ability, but can also be read as a negative stat when looking at all-around ability (why aren't Ovechkin, Malkin, Kovalchuk, and Max Afinogenov playing more shifts per game, and more PK shifts per game?)

Much like the trouble that Bill James and his peers had (and still have) objectively judging/calculating a baseball player's value in the field, I think that hockey's unanswered statistical questions lie within a player's defensive ability.

We can use the good ol' "eyeball test" to see that Henrik Zetterberg was a defensive terror against Crosby in two consecutive Cup finals series. We can see the hustle I player like Parise gives on the backcheck on every shift. We can even measure that Mike Modano is excellent at making his opponent turn the puck over.

But we don't really know how to translate all the things we see to numbers, and how to correctly weight the numbers we have to determine defensive value.

It stands to reason that guys like Datsyuk or Parise have a lower PPS because their defensive abilities draw them more shifts per game, as well as more shifts against top offensive opposition because their coaches know they can shut it down.

That doesn't mean that these players would score at the same PPS Ovechkin would. It just goes to show that the difficult-to-measure reasons behind a player's PPS may show why a higher PPS is good for one player, but that a lower PPS for another player may actually be an indication of the player's defensive commitment, ability, and efforts.

Shiv said...

This method is flawed. Points Per Shift overlooks one KEY fact: How long is each shift? Ovechkin and Kovalchuk may be well ahead of the pack in PPS but they also have one thing in common: They have longer shifts that almost ALL forwards.

A better alternative is to analyze Points Per Minute of Ice Time. This would allow you to determine who produces points most efficiently. But this has further implications. A player like Datsyuk will have a lower Points per Minute than Crosby if you combine the ENTIRE Time on Ice. Instead the analysis should be Points per Minute of EVEN STRENGTH and Points per Minute of POWERPLAY TIME.

Bendy said...

Right on, Nathan. PPS rewards players on the PP, & punishes those on the PK. Much like +/- does.

However, PPS may identify players who deserve more shifts, or younger players about to break out.

The Colonel said...

Good points by all. It is possible to weigh any stat with a good Excel spreadsheet. In fact, after I talk defensemen and rookies tomorrow, I am going to take a look at weighted PPS - how would it be projected for this same data group if everyone had the same number of shifts.

It does all come down to that question of how do you reward the player for doing what they do the most - operate without the puck on their stick. A great point was brought up on XM Radio a week or so ago where one of the guys said he asked his wife in the last 55 second power play the watched Ovechkin, how many seconds did Ovechkin have the puck? Her answer was 20 or 30 seconds. He timed it as 9 seconds, to include giving him the full second for the slapshot that scored. Do we leave that measure to the coaches and scouts since fans, even very Hockey-savvy ones from years of personal experience, tend to follow the puck and where it goes unless they are looking for something else in particular? It is easy to argue the 1980's Gretzky curl once across the blue line that went to Coffey and Kurri before Anderson slammed it home and 'robbed' the Great one of a point was nonetheless a contribution to that score.

So why did I choose what I have? Measuring the defensive effort and/or away-from-the-puck-time is very difficult, so for me it comes down to Win = success; the most scores = Win; and up to two assists + one goal = a score. So with deference to those who make their living away from the puck, I started with points and went on from there.

Oh, and I did think of using PPM (although I like PPN!). Your decimals get even smaller for one and I thought I was pushing my luck to be able to discuss the 'success' of the 'point-oh-four.' I chose PPG/Shifts Per Game = PPS because I figured the point a guy gets for coming over the boards and tapping a pass to a guy who scores 6 seconds into his shift that comes off the ice right away will be weighted out evenly by the 45 second shift in which he gets no points, which happens much more often than not.

And despite what one comment said on Kuklas Korner, I am a Poly Sci major who happens to think numbers can represent something, but they cannot define the euphoria of scoring the goal, or being the fan in your seats at the game who cannot help but leap up when that marker is tallied. :-)

Skeeter said...


+/- is not awarded during the PP or PK, so I'm not sure why you would say it rewards players on the PP and punishes those on the PK.

As for PPS, it does seem flawed. As Shiv mentions, Ovechkin and Kovalchuk have significantly longer shifts than other players. A good point of comparison would be Ovechkin and H. Sedin: Ovechkin plays an average of 21:03 minutes per game with an average of 19.9 shifts per game. H. Sedin plays an average of 18:53 per game with 24.2 shifts per game! In addition, Sedin averages almost a minute of PK time per game.

The Colonel said...

You have a greater chance of scoring a point each and every time you skate.

If you score 2 goals in 10 shifts, you are a .2 player at Point Per Shift.

If you measure it as 2 goals in (10 shifts at 45 seconds each, or) 7.5 minutes, you are a .267 player at Points Per Minute.

I suppose you can pick your poison there.

In further thought of accounting for defensive/away from the play tasks for a ranking, I've got an idea for that I will write up for next week.

Anonymous said...

Also note how the top 4 are Russians. The Gold Medal final against Canada is going to be interesting.

Steve said...

Behind the Net has extensive stats based on time on ice, including 5 on 5, 5 on 4, and 4 on 5 stats. Looking at goals/points per 60 minutes of 5 on 5 is more accurate than looking per shift. Here's the current list, sorted by goals/60 min of 5 on 5 time, with at least 5 games played:

Steve said...

Oops, my last post didn't have the link as a clickable link. Here it is again:


Brock said...

I have to say, I'm with the others who think a lack of consideration for shift length makes this method hugely flawed. Imagine if Ovechkin refused to come off the ice for entire periods! He'd be useless for much of the time, and a defensive atrocity, coasting around, but you can bet his PPS would be huge.

I understand the concern about using numbers that were two small when looking at PPM (points per minute), but that's why the statistic people use is P/60M. It's already out there and it's a much better measure of what you're analyzing here.