Win Probability Added (WPA) - playoff point scoring

[Bear of Bad News]

To the extent that this analysis suffers, it's because it ascribes all changes in win probability to scoring. Which is true (and probably biases towards eras where top-line players score more), but it's a broader impact that's being addressed above.

 

[overpass]

To the extent that this analysis suffers, it's because it ascribes all changes in win probability to scoring. Which is true (and probably biases towards eras where top-line players score more), but it's a broader impact that's being addressed above.

WPA (points) is subject to all the limitations of the points scored metric. And I agree it's more favourable towards eras where top-line players score more.

But so is every metric based on points or adjusted points. Raw point totals, scoring-level adjusted points, and ratio of points to league leaders all favour players with more ice time, players with more power play ice time, players who play on a stacked line with the best offensive talent rather than a more balanced line setup, players on teams with one scoring line rather than two or three, players on teams that rarely play their fourth line, and so on.

I don't know if there's a statistical way to adjust for the above factors that gets us to one number. But it's certainly important to consider these factors in any player evaluation.

 

[Bear of Bad News]

Agreed 100% - and I think you've been very transparent about that in the thread. It's a limitation of the data available, not necessarily of the approach.

The fact that each goal changes win probabilities more in low-scoring environments is not a limitation of the approach.

When the 1985 Oilers score eight goals in a game, they still only get one win as a result.

 

[overpass]

If I understand this analysis correctly (I might not), I think the reason 3rd-stage Gretzky looks more clutch is simply because scoring got lower and Gretzky's teams were less dominant.

Yeah that's most of it. Maybe all of it, although Gretzky did score a lot of clutch points in the 90s and I'm not sure if it was just because he was in a lot of close games.

The benefit of this framework is that it does account for the fact that goals in low scoring eras are worth less than goals in high scoring eras.

It may indeed underrate players on dominant teams that had more talent than they needed to win. Gretzky can't score as many go-ahead goals when Messier and Anderson are scoring them at a high rate.

 

[Bear of Bad News]

The benefit of this framework is that it does account for the fact that goals in low scoring eras are worth less than goals in high scoring eras.

Flip this one, though.

 

[overpass]

Flip this one, though.

D'oh! I'll just point to your post next time, lol.