major major
Registered User
- Feb 18, 2013
- 14,598
- 1,669
You guys have no idea how much I love a good statistical argument.
1. "Advanced Stat trickery". Bus, is that you?
It's P/60. Points divided by ice time, normalized to 60 minutes. It's the exact inverse of PROD, except it's 5 on 5 only. There's nothing advanced about it.
2. You can't use PROD because it gets skewed by volatile powerplay performance. To illustrate the volatility, right now Jason Chimera is a top 90 scorer according to PROD, largely based on the powerplay. And because it doesn't separate 5 on 5 from 5 on 4 and 4 on 5, it gives a huge advantage to powerplay players over penalty killers. Our penalty killers (like Cam) don't stand a chance in PROD. Perhaps for that reason, P/60 has succeeded it as the statistic of choice.
3. There's several different ways coaches use their best players. You can find Kuznetsov's on the second line and Couture on the third line, and neither getting 18 minutes. Or you can find Brian Gionta and Brandon Sutter above that bar.
4. No, you don't necessarily get more favorable match ups if you play less than 18 minutes. On a lot of clubs it's the third liners who get the hardest matchups.
Let's have a look at all of the conditions you just put on what qualifies one to be a top line player.
1) 18 minutes a night. That's about 80 forwards
2) Top 90 PROD. That's a high turnover group of 90.
3) Hardest matchups.
The overlap between those three gives you, what, 30 or 40 guys? And too much turnover. The Jackets might have no players in that group today and 5 in it a few months later. It doesn't tell you as much about the quality of players on your team.
Our sources are the same. Stats.hockeyanalysis is the source data for the hero charts. The only difference between hero P/60 and the source data, is that the hero charts are using a 3 year average. You're trying to refute my 3 year data by using data with a smaller sample size. That doesn't fly. It might make sense if a player like Hartnell is slipping, because he's exiting his prime, or if a young player suddenly emerges. But normally a 3 year average has more predictive power than the current year's performance.
Foligno probably averages about 18 minutes per game over the last three years. I'm not huge on ice-time as a measure of quality, but it still tells you a lot more about his value than the number of his line on the lineup card. We have a convention of numbering our lines here that is not based on in-game usage.
Hartnell is producing like a borderline top line player, but he's really not because of his ice time. This season Dubinsky isn't. I don't need whatever tool you produced when I can simply look at their PROD. What you produced looks like some advanced stat trickery. Nice tool, but I still think it's too heavily relied upon.
1. "Advanced Stat trickery". Bus, is that you?
It's P/60. Points divided by ice time, normalized to 60 minutes. It's the exact inverse of PROD, except it's 5 on 5 only. There's nothing advanced about it.
2. You can't use PROD because it gets skewed by volatile powerplay performance. To illustrate the volatility, right now Jason Chimera is a top 90 scorer according to PROD, largely based on the powerplay. And because it doesn't separate 5 on 5 from 5 on 4 and 4 on 5, it gives a huge advantage to powerplay players over penalty killers. Our penalty killers (like Cam) don't stand a chance in PROD. Perhaps for that reason, P/60 has succeeded it as the statistic of choice.
If you aren't playing 18+ minutes a night you really can't say you are a top line player. You are getting more favorable match ups.
3. There's several different ways coaches use their best players. You can find Kuznetsov's on the second line and Couture on the third line, and neither getting 18 minutes. Or you can find Brian Gionta and Brandon Sutter above that bar.
4. No, you don't necessarily get more favorable match ups if you play less than 18 minutes. On a lot of clubs it's the third liners who get the hardest matchups.
Let's have a look at all of the conditions you just put on what qualifies one to be a top line player.
1) 18 minutes a night. That's about 80 forwards
2) Top 90 PROD. That's a high turnover group of 90.
3) Hardest matchups.
The overlap between those three gives you, what, 30 or 40 guys? And too much turnover. The Jackets might have no players in that group today and 5 in it a few months later. It doesn't tell you as much about the quality of players on your team.
Actually, not.
In points per 60/5v5 (500 minutes minimum played) Hartnell ranks 18th in the NHL at 2.16. That's upper echelon this season. But, he often times hasn't faced first line competition as he hasn't been on the first line all that regularly this season.
Foligno is at 1.48 which ranks 123. Clearly outside of first line performance 5v5.
http://stats.hockeyanalysis.com/rat...ards&minutes=500&disp=1&sort=PCT&sortdir=DESC
Saad is 48th, Dubinsky at 83rd and Atkinson at 86th. Jenner is at 103.
Our sources are the same. Stats.hockeyanalysis is the source data for the hero charts. The only difference between hero P/60 and the source data, is that the hero charts are using a 3 year average. You're trying to refute my 3 year data by using data with a smaller sample size. That doesn't fly. It might make sense if a player like Hartnell is slipping, because he's exiting his prime, or if a young player suddenly emerges. But normally a 3 year average has more predictive power than the current year's performance.
Especially when one considers that he has been slotted as low as the 3rd line on a regular basis.
Foligno probably averages about 18 minutes per game over the last three years. I'm not huge on ice-time as a measure of quality, but it still tells you a lot more about his value than the number of his line on the lineup card. We have a convention of numbering our lines here that is not based on in-game usage.