lebdafor norris
sleepy
- Mar 14, 2011
- 3,828
- 889
Secondary assists - new studyWhy are you looking at McDavid's D+1-4 and comparing it to Matthews' D+0-3, especially when that's not either's pre-signing period? What McDavid did between 2016-2019 is irrelevant. What matters is what both players did during their respective pre-signing periods, and I have provided that for you, and even adjusted it based on league average scoring rates, at your request. Nothing changes the fact that Matthews was comparable to McDavid at time of signing.
Yes, he was, as I showed. He was the better primary point producer at both ES and PP, regardless of whether or not we adjust for league scoring rates. He was also trending as a generational goal scorer, and still is.
No, the adjustments I made were not a "gross understatement" of the reality. They are way, way more accurate than looking at numbers of players above some arbitrary threshold, which is literally useless.
The reality is, the difference from the things you said we needed to account for like equipment changes, were the same impact for everybody, but that impact was minimal; way less than people around here seem to think. League scoring rates didn't actually change all that much. As explained, more players in the top-end bracket is not a result of league scoring changes; they are a result of a mix of internal factors, such as an influx of talent, and evolving coaching behaviours, like line composition and increased TOI for stars (which is why it's so apparent in top-end raw points).
You're basically asking current stars to be paid less because current stars are taking on more responsibility than past stars. That's so backwards I don't even know what to say...
If that was true, he'd be getting more primary points, not less primary and more secondary... They had very comparable on-ice GF/60 at both ES and PP, especially when we consider that Matthews had to deal with split PP units for 80% of his pre-signing sample.
Highly debatable. Most studies on the "value" of a secondary assist place it somewhere around 5 times less valuable than a goal/primary assist, and they're not nearly as repeatable.
Nobody said primary points are everything. That's why I'm saying Matthews was comparable to McDavid at that point in time, not Matthews was way better than McDavid. But they are also not nothing. They are part of the picture, just like points.
As for Crosby and Perry, Perry was quite good back then at ES. He also played with Getzlaf, who placed just ahead of him in ES P/60, which boosted both of their ES totals relative to Crosby, who didn't play with anybody like that. Crosby was also just coming off his years of major injury/concussion problems, and had barely played in the previous 2 seasons. Crosby was also better on the PP, which again, you keep leaving out...
PP Primary points/60, 2012-2015
Crosby: 4.08
Perry: 2.62
Overall, Crosby was still the better point and primary point producer during those years.
One of the main points of the the p/60 and primary points stat is to determine the contribution and future production of certain players. The model that includes the both SA and primary was a FAR better at predicting a team's actual goals performance than primary points alone so I call BS on your devaluing of the SA. Note that a difference between a a R2 of .75 and .8 is a whopping 10% standard deviation and here we got a .72 and .82 R2 difference in favour of total points. The statistician Henseler, J., Ringle, C., and Sinkovics, R proposed that as a rule of thumb .75 should be the minimum amount for a "substantial" correlation, in other words point totals is a vastly superior tool to used than primary points alone when evaluating player contributions and future productions and using primary points alone doesn't even net (or barely meets depending on who you ask) a "substantial correlation."
Speaking of future production, have you ever stopped and think how awful your cherry picked p/60 stat at predicting actual future results? The fact that simply evalauting each players p/60 relative to their respective season has proven to be a much, much better indicator and predictor compare to you cherry picking 2 different time periods in order to make AM look better just proves how faulty your stat really is.
Anyway I think this discussion has run its course, we'll just have to agree to disagree.
Last edited: