Source for Similarity Scores of Current NBA Players?

[CappieP]

Is there a place where I can look up the similarity scores for current players to one another? I don't even know if similarity scores is the right term as I know that's been use to compare and project career arcs and I want to know about contemporaries.

But what I was wondering if there was a way to find out which groups of players have statistical profiles that resemble each other. I can think "small scoring point guards" and think of a set and eyeball their stats, but I'd rather be able to put in "Isiah Thomas" and get a read using someone else's recipe of which current players profile similarly to him. As long as the source includes the recipe of how these scores get cooked up, I would love to compare my eye ball sense of this to a statistical sense.

Links?

[Crow]

MikeG has done threads like this viewtopic.php?f=2&t=9073&hilit=similar
I thought there was one where the similar were restricted to current season. There are Mike G lists by player for his metric alone and a lists for similar for a blend of metrics I believe. Maybe he can link them or you can find it.

Did you try a search like this? search.php?keywords=similar&terms=all&a ... mit=Search If folks tried a couple different searches and browsed, they could find a lot of good stuff.

Here is another specific link viewtopic.php?f=2&t=8276&hilit=similarity

Also search for Ed Kupfer / factor analysis, David Sparks, Shoene... Ed Kupfer had a nice excel spreadsheet that was was to use in these vein maybe 8-10 years ago.

Or do a query of BR using Play Index / player season / advanced stats display for positions or player types using demographic and productivity criteria, copy into excel and setup a filtering system of your own using your preferred metric or blend of metrics.

[Mike G]

Are you looking for this kind of result?

Code: Select all

diff   per-36 rates    Sco   Reb   Ast    PF   Stl   TO   Blk
.00   Isaiah Thomas   24.5   3.2   6.9   2.3   1.2   3.0   .1
.16   Reggie Jackson  22.2   3.9   7.2   2.8    .9   3.4   .2
.17   Kyrie Irving    23.3   3.6   5.7   2.3   1.2   2.7   .4
.17   Damian Lillard  24.6   4.1   6.7   2.2    .9   3.3   .4
.35   Jeff Teague     20.2   3.3   7.5   2.7   1.5   3.5   .4
.45   Jrue Holiday    19.7   3.8   7.0   2.8   1.7   3.2   .4

.45   C.J. McCollum   21.0   3.4   4.6   2.4   1.3   2.6   .3
.50   Kyle Lowry      21.5   4.9   6.8   2.7   2.0   2.9   .4
.53   DeMar DeRozan   24.4   4.8   4.5   2.2   1.1   2.3   .3
.56   Dennis Schroder 18.7   4.0   7.4   2.9   1.5   3.8   .2
.57   Dwyane Wade     23.2   5.0   5.5   1.9   1.3   3.3   .7
.59   Kemba Walker    21.4   4.5   5.6   1.4   1.6   2.2   .4

These aren't "straight" per-36 rates but adjusted for lots of factors.
With the spreadsheet, though, you could use whatever numbers you want.

It's a Euclidean distance thing -- (sqrt(Sco1)-sqrt(Sco2))^2 -- for each of the columns shown. All then summed.
A diff of 1.00 is roughly tantamount to two players' position being different by 1 (i.e., a SF vs a SG)

EDIT: Just realized your example was "Isiah Thomas", which is how the HOF Pistons guard spelled his name.
Isiah's career rates vs 2016-only rates:

Code: Select all

diff   per-36 rates    Sco   Reb   Ast    PF   Stl   TO   Blk
.00   Isiah Thomas    18.4   3.7   8.8   3.1   2.0   3.8   .3

.12   Dennis Schroder 18.7   4.0   7.4   2.9   1.5   3.8   .2
.17   Jeff Teague     20.2   3.3   7.5   2.7   1.5   3.5   .4
.17   Jrue Holiday    19.7   3.8   7.0   2.8   1.7   3.2   .4
.37   John Wall       19.2   5.0   9.9   2.1   1.9   4.2   .8
.42   Eric Bledsoe    20.1   4.3   6.1   2.6   2.2   3.8   .7

.43   Kyle Lowry      21.5   4.9   6.8   2.7   2.0   2.9   .4
.45   Tyreke Evans    16.8   6.1   7.2   3.1   1.5   3.5   .4
.46   Chris Paul      21.9   4.4  11.7   2.8   2.3   2.9   .2
.47   Manu Ginobili   17.9   4.5   5.5   2.9   1.9   2.9   .3
.48   Goran Dragic    16.1   4.4   6.5   2.9   1.1   2.9   .3

This requires copy-paste from one spreadsheet to another. But a sheet could be built just for the purpose.
Isaiah Thomas, diff = .80, is only 20th most similar to Isiah.

[Kevin Pelton]

FiveThirtyEight's CARMELO projections have the top 10 most similar players:
http://projects.fivethirtyeight.com/car ... n-afflalo/

[Crow]

Didn't think about Carmelo when I responded. List of similarities likely to have changed some with move from RPM / BPM mix to solely BPM.

[Mike G]

I think the OP asked about current players; 538 is comparing current players to past/anytime players.
Maybe pickin's are too slim, and he's lost interest.

[DSMok1]

Didn't think about Carmelo when I responded. List of similarities likely to have changed some with move from RPM / BPM mix to solely BPM.

Ha, I totally missed that change! But Nate Silver's threatening with "Frankly, we have designs on our own plus-minus metric that would eventually displace both BPM and RPM, but that’s something that will have to wait for a future incarnation of CARMELO." http://fivethirtyeight.com/features/wha ... r-2016-17/

Neil, what's he up to?

[sndesai1]

my initial idea would be to do a weighted euclidean distance of the variables used in a statistical +/- (probably excluding minutes per game) and the variables used in a player tracking +/- with the weights being the absolute value of those variables' coefficients...maybe also figure out a way to include height/weight


in terms of existing tools, carmelo is the only thing i can think of, but i'm not that familiar with the goal of their similarity score. are they looking for similarity of player style? or similarity of career arc? or some mix?