deadhead
Registered User
- Feb 26, 2014
- 49,215
- 21,617
No single stat is perfect. Game score tells us Jake is a solid 2nd liner at 5v5, his production says 1st liner. There are not many guys in the NHL that all of GAR, GS and scoring itself will tell you are actual 1st liners, even if alone each says around 90.
There are like ~15-20 guys in the NHL who put up great scoring, relative underlying stats and raw underlying stats consistently.
Case in point, over the last 3 seasons here are the players over "1st line level" in all the following areas:
1.9 P/60
+2.00 CF Rel
+2.45 ExGF Rel
Leaves you with 30 players... without the raw stats included.
If you want to be a top team, you can't have a 1st line composed of players in the top 90, a 2nd line from 91-180 and so forth.
A team composed of average players in each tranche would be an average team.
You want to be in the top third or so in all areas, obviously, there will be some divergence, maybe you have an elite 1st line, or elite top D-pair.
Or you lack those elite players but have a 3rd line better than most team's 2nd line and beat them with depth.
But using that as a baseline, I'd say:
1st line players should be in the 1-45 group, totaling say 90 or so (i.e. 10th, 35th, 45th)
2nd line from 46-135
3rd line from 136-225 or so.
In other words, you don't want players on any line who are below average league wide for that line.
We have crude metrics for defense (baseball has the same issue in calculating WAR). xGF% and EV GAR try to account for it, not perfect measures, which is why you want to use multiple measures. But you want to try and account for it.
Scoring does matter, and goal scoring more than assists (there's about 2.25 points per goal, but one goal per goal).
But defense also matters. Most scorers tend to be below average to average defensively (self-selection, scorers are more likely to get PT with below average defense than nonscorers), so those who are above defensively should be ranked higher than their point totals.
PP scoring matters, but more of a unit measure since individual PP scoring for most players is primarily determined by opportunity, not skill (which is why PP GAR is a good check on raw PP scoring statistics).