Earlier, Jose Martinez introduced the MagicMetric rating system to apbr after tracking me down for the details. I subsequently compared the magic metric to the EFF rating used on hoopstat.com in a paper posted to this group. Incorrectly, I attributed this compairson to John Hollinger's PER rating, in the comments.
To make up for this error, I prepared another comparison available here:
http://www.upct.es/~beside/Textos/MAGIC ... gCards.pdf
The general problem of comparing rating systems is not simple to solve. I think the playing card approch might be interesting to the readers of the group. Comments are most welcome.
A playing card approach to evaluating rating systems.
Re: A olaying card approach to evaluating rating systems.
There's no doubt Chris Paul is better than both Kobe and Harden, and probably Durant in my opinion, but that's very much up for debate. Lopez is misplaced by PER, but then Melo shouldn't be being considered a top twelve player even by box-score metrics.
In the end though reliance on the smell/eye test isn't good, as there's a big bias towards 'pretty' players. Simple out-of-sample testing would best reveal the most accurate metric.
In the end though reliance on the smell/eye test isn't good, as there's a big bias towards 'pretty' players. Simple out-of-sample testing would best reveal the most accurate metric.
Re: A olaying card approach to evaluating rating systems.
He may be better than either, but he's not better than both.There's no doubt Chris Paul is better than both Kobe and Harden...
I've got Melo at #9 in both eWins and eWins/minute.... Melo shouldn't be being considered a top twelve player even by box-score metrics.
He's one of the top 3 scorers in the game, and he does a few other things.
There are just 25 other player-seasons since 1971 with: 28 PPG, 14.5 Ast%, 9.5 Reb%, and a TS% over .555 -- Jordan did it 7 times, LeBron 4, Bird 3; Kareem, Robinson, Shaq, and Durant, twice each.
English '83, 'Nique 93, and McGrady '03, in their career years.
http://bkref.com/tiny/AIqnD
Re: A olaying card approach to evaluating rating systems.
Perhaps not together, but individually... 
Melo is having a career year, but the Knicks haven't struggled without him, and he's been out for decent stretches. He may be a top 'scorer', but he's not in the same league as LeBron or Durant, Melo needs 34% usage to be a top scorer, he's still taking (optional) bad shots, and doesn't assist enough given the attention he draws. His D isn't good, it's actually quite poor. He's a star player, he's just not top twelve in the league.
Melo is having a career year, but the Knicks haven't struggled without him, and he's been out for decent stretches. He may be a top 'scorer', but he's not in the same league as LeBron or Durant, Melo needs 34% usage to be a top scorer, he's still taking (optional) bad shots, and doesn't assist enough given the attention he draws. His D isn't good, it's actually quite poor. He's a star player, he's just not top twelve in the league.
Re: A olaying card approach to evaluating rating systems.
I wish I had posted this when I was able to spell, lol. I think Melo is top twelve this year, but everyone ie entitled to their own opinion.