Quarter Season Statistical Analysis of Top 20 Point Producers

[Avelanche]

should be noted with Mackinnon he was shooting like 7% before rantanen, last season he shot ~14%. so mack is not over achieving or something. he has a great shot on a bad possession team.

 

[FalcorMulch]

I've never seen people get so angry at data before. No one is asking you guys to take this as gospel. It's a simple but interesting analysis of how guys are playing relative to career norms and what it could mean for their season point totals.

 

[FalcorMulch]

I will say that as someone who works with statistics, it all really depends on what analysis you choose to apply to a dataset. Two statisticians can process the same data set in two very different methods and end up with vastly different results. As subjective as the "eye test" is, statistical analysis can be just as dependent on the individual judgments of the statistician. Not that stats are useless, far from it but while indeed, "numbers don't lie" it is more accurate to say "numbers don't lie, but they can be processed and interpreted to say whatever you want them to".

You're right but the benefit of numbers is that the analyst's process, judgment, and interpretation is usually laid out for everyone to see and evaluate (at least if the analyst doesn't want their work to be dismissed). People can then determine for themselves, to a degree, if the analysis makes sense and actually supports the conclusion it's claiming to support. On the other hand, it's very difficult to evaluate what someone sees with the eye test.

 

[stokes84]

There are four other guys with 24 points that you just arbitrarily cut out. What’s the deal with that?

 

[GirardSpinorama]

oiSH% is only calculated at even strength. All of the numbers in the table use even strength data only. And of course the point totals include both EV and PP points so the math isn't exact but the data for PP points is not as easily available and I didn't feel like doing the extra work.

Theres another thread that stated pp is at the highest success rate levels since the 80, so that info needs to be accounted for or this analysis has major flaws.

 

[TDK67]

We are at the quarter mark of the season and it seems like the highest scoring season in a long time. We have 50 players producing at PPG pace or better! So let's take a look at the top 20 scorers (since I didn't feel like doing the top 50) and compare their up-to-date production against their career production and see which players are likely going to sustain their pace, improve it or see it degrade over the course of the season.

I have used ioSH% (on-ice shooting percentage) as the statistical measure. This is the percentage of shots that result in goals while a player is on the ice. In the table below, I have listed the current oiSH%, career oiSH%, games played, points so far, points predicted by the career oiSH% (that is, how many points a player would have if they had their career oiSH% this season), the difference in points, predicted PPG based on points a player would have if they had their career oiSH% and finally the predicted point totals.

Player Name	oiSH%	Career oiSH%	GP	Points	Points based on career oiSH%	Delta	Predicted PPG	Predicted Season ending points (predicted PPG*82)
Rantanen	14.5	9.7	20	32	24	-8	1.2	98
MacKinnon	13.5	9.7	20	29	21	-8	1.05	86
Marner	13.4	9.8	21	28	20	-8	0.95	78
McDavid	9.6	10.5	20	28	31	+3	1.55	127
Malkin	11.6	10.3	19	27	24	-3	1.26	104
Duchene	15.5	9.3	21	27	16	-11	0.76	62
Bergeron	12.8	8.8	19	26	18	-8	0.95	77*
Point	11.8	10.2	21	26	22	-4	1.05	86
Rielly	14.2	8.6	21	26	16	-10	0.76	62
M.Tkachuk	10.1	8.5	21	25	21	-4	1.0	82
Tavares	13.7	9.3	21	25	17	-8	0.81	66
Pastrnak	11.4	9.6	20	25	21	-4	1.05	86
P. Kane	10.9	9.6	20	25	22	-3	1.1	90
Domi	12.8	10.2	21	25	20	-5	0.95	78
Stone	13.4	10.5	21	24	19	-5	0.9	74
Monahan	11.1	9.9	21	24	21	-3	1.0	82
Kucherov	10.1	10.1	21	24	24	0	1.14	94
Kessel	12.9	9.5	19	24	18	-6	0.95	78
Giroux	10.8	8.9	20	24	20	-4	1.0	82
Gaudreau	10.7	10.5	21	24	24	0	1.14	94

[TBODY] [/TBODY]

It's interesting to note that all but three of the top producers are overachieving. At or below their career oiSH are McDavid, Kucherov and Gaudreau.

Also noteworthy is the range of career oiSH%. Lowest is 8.5 and highest is 10.5. Compare this range to the range of year-to-date oiSH% with lowest of 9.6 and highest of 15.5. You notice right away the much bigger range (5.9 vs 2 in the career oiSH%), which is explainable by the fact that we only have 20 games of sample size and numbers normalise over the course of the full season.

Top 10 predicted scorers are
McDavid- 127
Malkin - 104
Rantanen - 98
Kucherov - 94
Gaudreau - 94
Kane - 90
MacKinnon - 86
Point - 86
Pastrnak - 86
Tkachuk, Monahan and Giroux tied at 82

The one outlier in the data set is Duchene, who is having a lot of luck early on and is projected to finish the year with 67 points, a far cry from his current 100+ point pace.

Another interesting prediction was Tavares's 66 points. That really surprised me but the numbers are telling us that he really should have more points right now given his very high oiSH%. Not sure what to make of it. I would bet he finishes with more than 66 points, though

I did not look into scorers below the top 20 and I'm sure some of them will pick it up and end up in the top 20 or even top 10 in scoring.

*Obviously Bergeron will not play 82 games.

Although individual shot% is part of oISH%, it seems shortsighted to completely ignore that for each top producer in this small sample. Most of the top point getters are scoring goals at a far higher % than their career shot% which is worth considering as those numbers are likely to impact individual production just as much (or more) via regression as general oISH%. Another way to look at it is that we're clearly not getting 20 something different players finishing with 40-50+ goals this year.

A more accurate and thorough analysis would separate 5v5 vs PP numbers, account for ice time, and regress shot% and oISH% separately.

 

[hatterson]

One thing I'll note is that predicted points shouldn't be 82*predictedPPG but rather currentPoints+(gamesRemaining*predictedRemianingPPG)

Rielly shouldn't be expected to maintain his current 1.23 ppg, but he's not going to give back the 10 points above "expected" that he is currently.

Edit: Essentially that means for this chart you can take the Predicted Season Ending Points column and add it to the Delta to get the "If from here on out the player played at 'expected' level, here's what they'd finish with"

 

[Rich Nixon]

Frankly, I think this is absolutely silly. It's neither a good stat nor a good statistical analysis, and the concept is bad. So Duchene SHOULD have way less points based on career averages? Well, he doesn't.

This isn't how anything works, really. oiSH% is a more interesting stat to me than most advanced stats, but it still suffers from the problem of being a stat in a sport that isn't conducive to statistical analysis.

Not to mention the career number you used for Giroux (8.9%) is...wrong. It's 10.4%, according to hockey-reference.com.

 

[42]

One thing I'll note is that predicted points shouldn't be 82*predictedPPG but rather currentPoints+(gamesRemaining*predictedRemianingPPG)

Rielly shouldn't be expected to maintain his current 1.23 ppg, but he's not going to give back the 10 points above "expected" that he is currently.

Edit: Essentially that means for this chart you can take the Predicted Season Ending Points column and add it to the Delta to get the "If from here on out the player played at 'expected' level, here's what they'd finish with"

EDIT: Initially, I agreed with this but then I realized that my calculation is correct. No, Rielly won't give back the points but the expectation here is that by the end of the season, each player will regress to their career average so that the total should be calculated over the full season. The regression means that players will shoot lower than their average and therefore get fewer point.

 

[42]

Frankly, I think this is absolutely silly. It's neither a good stat nor a good statistical analysis, and the concept is bad. So Duchene SHOULD have way less points based on career averages? Well, he doesn't.

This isn't how anything works, really. oiSH% is a more interesting stat to me than most advanced stats, but it still suffers from the problem of being a stat in a sport that isn't conducive to statistical analysis.

Not to mention the career number you used for Giroux (8.9%) is...wrong. It's 10.4%, according to hockey-reference.com.

Claude Giroux Stats | Hockey-Reference.com - look it up, it's 8.9%

 

[NoName]

I suspect Tavares' low predicted points are the result of him having low PP points compared to EV points (6 PP so far only). See my previous comments about a shortcoming of my analysis, which basically does not take into account the fact the oiSH% is calculated for EV production only but I used total points in my prediction. As I said, I might re-do the calc (if the motivation strikes me) to see what differences it produces. I suspect Tavares' number will go up.

For the record, full respect for putting the time in and trying something like this.

 

[Rich Nixon]

Claude Giroux Stats | Hockey-Reference.com - look it up, it's 8.9%

That's even strength only, my guy. Y'know, excluding the scenario where like 20+% of goals in the NHL are scored. Here's his all-situational stats: Claude Giroux NHL Advanced Statistics (All) | Hockey-Reference.com

Hoping you didn't do this with every player because...lol.

EDIT: You did. Hahahahaha. So this analysis is only worth the grain of salt it is if no penalties are called for the rest of the season.

EDIT EDIT: I see you did have an exchange about this previously in the thread under the assumption that the all-situations data isn't available. It is, all on the "Advanced" tab of hockey-reference for each guy. You can sort it by situation. Giroux's actually at 8.4% shorthanded, not bad! I actually didn't know you could sort it that distinctly until now. So, now we both know that!

Not tryna be mean or anything but...this is a bad concept to start and just completely worthless if it's even strength only. I respect the effort but I knew just looking at it that that your numbers were completely off somewhere, because there was no way in hell that Marner had a significantly better career oiSH% than Giroux.

 

[CDN24]

My prof used to say "bad data is worse than no data"

My stats prof was always quoting mark twain I think who said there are 3 kinds of lies.........

Lies, damn lies and statistics

 

[Tkachuky]

Marner will take Duchenes spot and Eichel takes Monahans

I won’t argue about Eichel with you, but Monahan was on pace for PPG for the better part of last year while playing with some bad injuries. It was not a fluke. the top line in a Calgary is even better than last year with the addition of Lindholm over Ferland.

 

[WhiskeyYerTheDevils]

I am consistently surprised by how underrated Johnny Gaudreau is on these boards. He's actually been playing "mediocre" hockey by his standards and is still on pace for 94 points with really no benefit of puck luck.

 

[Tage2Tuch]

I won’t argue about Eichel with you, but Monahan was on pace for PPG for the better part of last year while playing with some bad injuries. It was not a fluke. the top line in a Calgary is even better than last year with the addition of Lindholm over Ferland.

I wasn’t saying anything to discredit Monahan the flames or if they are better or not. I just said two players I thought would pass two players.

 

[Tkachuky]

I wasn’t saying anything to discredit Monahan the flames or if they are better or not. I just said two players I thought would pass two players.

Understood. I wasn’t pointing at that. Was just wanting to explain my reasoning as to why I see Monahan being at PPG.

 

[42]

There are four other guys with 24 points that you just arbitrarily cut out. What’s the deal with that?

I took the top 20 point producers as listed by hockey-reference at the time of writing. Some players with equal number of points as those at the bottom of the list were cut off. I did not look closely to see what the sorting algorithm hockey-reference uses to apply to players with equal number of points. I'd assume PPG is one criterion and there are likely secondary and tertiary conditions in play as well should PPG be equal.