Mike G wrote:Remind us what is EWA, and then what is 'faux EWA' ?
How do you conclude that there is "extra value brought in being special for a shorter period relative to being solid for longer, which cumulative metrics doesn't account for
?
I've honestly seen no evidence for this; nor can I imagine where such evidence is found.
If a player is a truly transcendent talent for a short time -- Bill Walton might be the only such example -- then maybe. But he was also key as a 6th-7th man for a Celtics championship.
Would 14 "solid" seasons have been a better career, or likely to involve more titles? More wins for more teams?
Penny Hardaway was great for a short time, got zero rings. How many such could've-been stories are there? Grant Hill is another.
EWA is a PER derived Hollinger stat. But cumulative rather than per minute.
For EWA (on a yearly basis) this millenium see
http://insider.espn.go.com/nba/hollinge ... fied/false
ESPN http://insider.espn.go.com/nba/hollinger/statistics/_/sort/VORPe wrote:VA: Value Added - the estimated number of points a player adds to a team’s season total above what a 'replacement player' (for instance, the 12th man on the roster) would produce. Value Added = ([Minutes * (PER - PRL)] / 67). PRL (Position Replacement Level) = 11.5 for power forwards, 11.0 for point guards, 10.6 for centers, 10.5 for shooting guards and small forwards
EWA: Estimated Wins Added - Value Added divided by 30, giving the estimated number of wins a player adds to a team’s season total above what a 'replacement player' would produce.
Faux EWA is how I'm reffering to Hpanic's (no longer online) PER based value added measure. Here's as much of the methodology as was given
hpanic7342 on RootZoo forums wrote:I wanted to use, as a template, a basic VORP formula, which, in its most general form, looks like this:
VORP = (PER - ReplacementLevel)*minutes
But I tweaked it in some ways that I felt were appropriate. Firstly, the formula wasn't quite so linear. I made an adjustment to the general form such that if two players had equivalent VORP (as stated above), but one had a higher PER (and those lower minutes), that player had a higher rating. I did this to mesh with my idea of what "greatness" is. I prefer someone like say, David Thompson to someone like Otis Thorpe.
I adjusted each player's PER to reflect his abilities on defense. The question I asked was, "By the standards of a great player, how good was this guy on defense?" If someone's defense was what we'd normally expect from someone who was really good, I left his PER alone. For guys who were just average or bad defenders, I knocked a point off. For very good defender, I gave 1 point. For transcendent, amazing, game changing defenders (Pippen, Olajuwon, Mutombo, Robinson, Frazier, etc.) I gave two points. Bill Russell got 4 points. A difference of even one point made a huge difference in where a player got ranked, so I REALLY gave Bill Russell the benefit of the doubt here.
Playoff PER and minutes were treated similarly, with two exceptions. The first is that I set the replacement level higher for the playoffs. It wasn't enough to just get to the playoffs a lot; players needed to play like stars once they were there in order to get credit for what they did. Those who did were rewarded handsomely, because playoff minutes were weighted three times more than regular season minutes.
We have to remember all of the implications of that last paragraph: Karl Malone's career playoff PER was 21.1. This is quite a bit lower than his regular season 23.9, but it's still damn good, and I'd rather have Karl Malone than almost anyone else, even in the playoffs. It's chopped liver however compared to Hakeem Olajuwon's 25.7, so this was a part of the formula where a guy like Olajuwon could pull ahead.
hpanic7342 on RootZoo forums wrote:Also, I need to mention: there was an adjustment made for when a player played. Basically, the earlier his career took place, the more his rating was scaled down. Players in the 50s didn't have to play against blacks, in the 60s they didn't have to play against Southern blacks, and not until the 2000s did they have to play against lots of internationals. I wanted the ratings to reflect that the competition's gotten better.
I call it faux EWA to make clear what it's lineage is, and to show it has been adjusted (including unclear replacement level values, era adjustments, defensive adjustments). Technically it's more faux VA, since the numbers produced are way bigger than win totals. But I think EWA is more recognisable and less generic.
As for the advantage of say a one year 30 PER player over two years 20 PER player (or 20 and 15), the reasoning is simple (those values are for illustrative purposes only and won't reflect an exact 2x difference). There are only so many minutes available, for team whose goal is to win a title, to get not just above average, but substantially above average (as winning a title will usually require). With the one great player your putting youself a significant distance above the pack, with additional minutes with which to deepen that advantage. With a longevity player for two years, you're spreading your goodness thinner. I don't know how/whether player metrics translate to team ones. But say that you could have a team of players giving a 2 point advantage per postion (over the average player) for one year, or a one point advantage for 2. The first team would post an SRS above 10. Of teams with a 10+ SRS, 5 of 6 have won a title (the sixth, the '72 Bucks had the misfortune of playing in the same year as the slightly more dominant '72 Lakers). By contrast 172 teams have posted a 5+ SRS. Even with two bites at the cherry, your chances of winning a title are still fairly slender.
Now there's a chance the team fail to use the extra minutes availed of them by further distancing themselves from the pack, they might even, on occasion fill them with bad players. Then you get something like T-Mac on the Magic. But in a league with a nominal 1/30 chance of winning (though greater in the past), and winning a title is your goal, I'd suggest (given the option) you need to concentrate your greatness.
And that's all leaving aside salary cap stuff (modern era individual player max means great players are by far the best value players), though this is fair enough as it is extrinsic to basketball ability.
Obviously there's a lot of technical issues about what measures to use (that would actually translate into wins) and what it would actually mean being, say, "double" another player (e.g. is it distance from the average, distance from replacement level etc). But the above is a general rationale for why you want short term greatness rather than longer term goodness.
post edited to correct formatting on quotes
Also I note that the faux EWA measure already did bump peak performance, though to what degree, we cannot be sure.
