P/60 stats: How useful and reliable are they? Does production increase linearly with ice time?

[Snippit]

Tangent from a different thread but an important debate to have.

P/60 stats are obviously useful to analyze players with differing amounts of ice time.

Obviously when you increase a player's ice time, they'll probably put up more points. The question is, does production increase linearly with ice time? That is, if a player in a season scores 70 points over a full season playing 16 minutes a night, how likely is it that they would have scored 88 points playing 20 minutes a night? (the same p/60).

Another way to put it: If I made a plot of ice time vs production, would that line be straight all the way for the first 22-23 minutes?

My instinct tells me that it's not likely, because 1) the smaller sample size of 16 minutes would be more prone to variation and 2) fatigue would come into play as well.

However I'm wondering if there's actually been research done into this. If the increase in production is linear with increase in ice time, then p/60 is a fantastic stat and we should use it instead of p/gp.

 

[VoluntaryDom]

i find p/60 to be useful

 

[CantLoseWithMatthews]

it's good for adding context, but it isn't outright perfect. it's pretty dumb to neglect stats for not being completely perfect though. For me, looking at raw numbers and ignoring everything else is stupid

it's also funny that you think p/g is linear and worthwhile but p/60 isn't. it's the exact same idea

 

[Super Hans]

Quality of competition plays a bigger role in P/60 than a player's fatigue of playing 4 more minutes a game.

 

[snipes]

It's a terribly misused stat by a lot of people.

However, there is some utility in it for judging bottom 6 players as part of an overall analysis for example.

It's when people espouse "better 5 v 5 ES primary points/60" is more important than a 30-40+ difference in raw point production that makes the stat look silly and extremely cherry picked.

 

[WetcoastOrca]

It’s used far too simplistically. People assume that points will just increase proportionally with more ice time.

 

[Isaac Nootin]

It’s used far too simplistically. People assume that points will just increase proportionally with more ice time.

It depends where those minutes come from. PP, PK, QOC...lots of factors.

 

[Snippit]

It’s used far too simplistically. People assume that points will just increase proportionally with more ice time.

I also think this is an unrealistic assumption but it's tough to prove.

One thing that could be done is to take a player with a wide spread in ice time over the course of a season, divide their GP into ice time intervals (GP with 16 minutes, GP with 17 minutes, GP with 18 minutes) then average out the production in each of those blocks and plot it on a line graph. Then you could see how linear the trend is.

Maybe @Canada4Gold has a better idea but I'd like to see more research done into this

 

[CantLoseWithMatthews]

It’s used far too simplistically. People assume that points will just increase proportionally with more ice time.

no they don't. P/60 is descriptive of what happened, not predictive of what will happen. People can assume players will get more points if they get more TOI, but that's common sense

 

[jetsforever]

PPG seems a bit unfair for guys with lower ice time, although P/60 isn't perfect either. Both need to be taken with context.

 

[McFlyingV]

Logic would dictate that at some point fatigue on a nightly, weekly, and monthly basis would become a factor with increasing minutes. It's impossible to know at what point more minutes decreases P/60, and I'm sure it depends heavily on the individual and their usage. I don't mind the stat if you're comparing players with similar ES usage as an example, but to think that say an extra 4 or 5 minutes per game is not going to have negative factors of fatigue is just ridiculous. More ice time means less time to recover between shifts, it means more fatigue when you're on the ice both physically, but also mentally. We know fatigue is detrimental to decision-making. Fatigue it also far more than a one-night thing as there are without a doubt cumulative effects over an entire season.

 

[Riseonfire]

Last years top 10 in p/60 (excluding super low sample size) for some reference.

Malkin
McKinnon
Hall
Marchand
Kucherov
McDavid
Giroux
Kessel
Stamkos
Barzal

 

[HugginThePost]

Depends on what your agenda is....

If you're using it to prove that your player is better than your rival teams player, it's perfectly acceptable.

We see it all the time in the Matthews vs McDavid pissing matches all the time.

As was mentioned above it's all dependent on how that extra time is utilized.

As with McDavid, he adds over a minute a game on the PK. His P/60 clearly aren't going up during that time.

Of course, I do believe McDavid does benefit from his increased ice time more than the average player. Anyone that watches him play knows he creates a shit ton of chances every time he's on the ice, more so than other player. Cut his ice time to 16 - 18 minutes a game and I don't think he wins the Art Ross, not with that supporting cast.

 

[zeke]

It’s used far too simplistically. People assume that points will just increase proportionally with more ice time.

nobody assumes that, actually.

though the anti-p60 crowd does seem to want to assume that p60 goes down with more usage, which isn't actually supported by anything.

using a random example like, say, Auston Matthews - his ES minutes increased significantly last year, as did his quality of competition.....and yet his p60 increased even more. in his case, more and tougher minutes resulted in better p60, not worse.

but the more important point here is to realize that while p60 has flaws, raw points is a stat with even bigger flaws.

 

[McFlyingV]

Last years top 10 in p/60 (excluding super low sample size) for some reference.

Malkin
McKinnon
Hall
Marchand
Kucherov
McDavid
Giroux
Kessel
Stamkos
Barzal

Not a great way to use P/60 in the first place when you don't remove PK time.

 

[Snippit]

nobody assumes that, actually.

though the anti-p60 crowd does seem to want to assume that p60 goes down with more usage, which isn't actually supported by anything.

using a random example like, say, Auston Matthews - his ES minutes increased significantly last year, as did his quality of competition.....and yet his p60 increased even more. in his case, more and tougher minutes resulted in better p60, not worse.

but the more important point here is to realize that while p60 has flaws, raw points is a stat with even bigger flaws.

How much of that do you attribute to player development though? I think we should only use players that are past their development years or use chunks of games from the same season.

P/60 staying the same with more usage seems unlikely to me, but again I don't think there's been much research into this. Have to control for development and the kind of ice time

 

[Hasa92]

It's a good stat but just like many other stats you need to look at it with context.

 

[Riseonfire]

Not a great way to use P/60 in the first place when you don't remove PK time.

Agreed. Thats why i posted the flat P/60 top 10 for reference. I'd like to see the top 10 list for all scenarios but don't have the time. Maybe you do?

 

[Deficient Mode]

They're extremely useful to whittle out some of the white noise from which teams have successful power plays in a particular year - something that is far more systems driven from year to year and independent of individual player talent - and which players are playing on their team's PP and which players aren't. The arguments against the use of per 60 stats are weak and rooted in cliches. The leaders in per 60 scoring at 5 on 5 correlate strongly with the leaders in overall points, and the guys playing the most minutes are also scoring the most per minute. It's not like the list of the top 50 guys in per minute scoring are a bunch of second and third line players. There is frankly no evidence at all that playing fewer minutes a game leads to a higher scoring rate for forwards, and a ton of evidence to the contrary. If a player has high per minute scoring over a sufficient (multi-year) sample, it is likely that they are a high end ES scorer, even if they don't put up point totals as impressive.

In addition, all available evidence suggests that there is a far higher variance in quality of teammate than quality of competition - another factor that prevents second and third line quality players from thriving in them, all talk of "they don't draw the toughest matchups" aside. The guys who are playing fewer minutes are almost certainly also playing with inferior teammates. While in a given year you may have a few second and third liners on great teams (Tampa, Boston) rate very highly in per 60 scoring, those will usually fade away over multiple years.

 

[biotk]

One thing that could be done is to take a player with a wide spread in ice time over the course of a season, divide their GP into ice time intervals (GP with 16 minutes, GP with 17 minutes, GP with 18 minutes) then average out the production in each of those blocks and plot it on a line graph. Then you could see how linear the trend is.

This would be extremely stupid. Most star players see their ice time go up when they are behind in a game - for instance when going up against a really hot goalie, while at the same time going up against a team that has shortened their bench in favour of their best defensively responsible players, playing as low risk hockey as they can.

 

[Snippit]

This would be extremely stupid. Most star players see their ice time go up when they are behind in a game - for instance when going up against a really hot goalie.

Harsh lmao but okay.

Doesn't seem like there's a good way of going about it unless you do it for the same player in the same season though.

 

[Snippit]

They're extremely useful to whittle out some of the white noise from which teams have successful power plays in a particular year - something that is far more systems driven from year to year and independent of individual player talent - and which players are playing on their team's PP and which players aren't. The arguments against the use of per 60 stats are weak and rooted in cliches. The leaders in per 60 scoring at 5 on 5 correlate strongly with the leaders in overall points, and the guys playing the most minutes are also scoring the most per minute. It's not like the list of the top 50 guys in per minute scoring are a bunch of second and third line players. There is frankly no evidence at all that playing fewer minutes a game leads to a higher scoring rate for forwards, and a ton of evidence to the contrary. If a player has high per minute scoring over a sufficient (multi-year) sample, it is likely that they are a high end ES scorer, even if they don't put up point totals as impressive.

In addition, all available evidence suggests that there is a far higher variance in quality of teammate than quality of competition - another factor that prevents second and third line quality players from thriving in them, all talk of "they don't draw the toughest matchups" aside. The guys who are playing fewer minutes are almost certainly also playing with inferior teammates. While in a given year you may have a few second and third liners on great teams (Tampa, Boston) rate very highly in per 60 scoring, those will usually fade away over multiple years.

What's the evidence?

That's what I've been asking for.

 

[Pierce Hawthorne]

It has its uses like any stat when applied properly.


Unfortunately people on this site do not know how to properly use statistics in the right context or how to properly pair stats with other stats to give any sort of meaningful value.


One statistic by itself is almost always meaningless. But on this site that's often all you see in arguments. Someone will cherry pick a single stat to benefit the agenda they're pushing.

 

[MessierII]

No it does not increase linearly with ice time which is the problem.

 

[biotk]

Harsh lmao but okay.

Doesn't seem like there's a good way of going about it unless you do it for the same player in the same season though.

I agree, that it is extremely difficult. I generally will only compare P/60 (my preference is 5v5 P1/60 or 5v5 P/60) with players that have similar QoC and QoT. I feel pretty confident in doing so because I am measuring two players in as similar situations as I can. That is not the case with raw points where some players are playing twice as much PP time on a completely loaded line and others are not. (It is also the case where I find 5v5 points, especially 5v5 primary points to be the most repeatable, whereas I find PP points to be almost completely unrepeatable from season to season).