Author Topic: Introducing the J Rating (or J Points)  (Read 2746 times)

0 Members and 1 Guest are viewing this topic.

Introducing the J Rating (or J Points)
« on: March 01, 2019, 04:01:57 PM »

Offline gift

  • NCE
  • Ray Allen
  • ***
  • Posts: 3983
  • Tommy Points: 291
For fun, I developed my own rating system for NBA players using box score statistics. This one-number performance rating takes into account traditional counting stats and balances them with shooting efficiency, and plus/minus in a novel way.

This rating system is at least as flawed, if not more flawed, than any other rating system ever created. But it was a fun project and I hope to continue tracking players year by year just to see what kind of numbers result over time.

It is not meant to determine how good a player is. It is aimed at answering the question: how good of a game or season did the player have? Since the best players will tend to have the best seasons, they will tend to be ranked higher in the J Rankings.

Because this rating takes into account and rewards cumulative stats as well as team success, while making no claims in regard to “best player”, it might be a good number to look at in consideration of the MVP award.

For each game a player appears in, they receive a J Rating number for that game. These numbers can be averaged to achieve their J Rating on the season. They can also be summed to create their total J Points on the year (though I’m not sure this is meaningful).

Here are the Boston Celtics J Ratings for the 2018-2019 season:

Player Name       Games Played   J Rating
Kyrie Irving               51              0.98
Jayson Tatum               62              0.73
Al Horford               52              0.71
Marcus Morris               57              0.65
Jaylen Brown               57              0.57
Gordon Hayward       56              0.57
Marcus Smart               61              0.53
Terry Rozier               60              0.49
Daniel Theis               48              0.47
Aron Baynes               34              0.45
Robert Williams       21              0.41
Brad Wanamaker       27              0.37
Semi Ojeleye               42              0.29
Guerschon Yabusele    30              0.23
PJ Dozier                         3              0.16

Re: Introducing the J Rating (or J Points)
« Reply #1 on: March 01, 2019, 05:03:28 PM »

Offline PhoSita

  • NCE
  • Robert Parish
  • *********************
  • Posts: 21835
  • Tommy Points: 2182
TP for the effort.

I'm curious what kind of math is behind the numbers.
You’ll have to excuse my lengthiness—the reason I dread writing letters is because I am so apt to get to slinging wisdom & forget to let up. Thus much precious time is lost.
- Mark Twain

Re: Introducing the J Rating (or J Points)
« Reply #2 on: March 01, 2019, 05:07:20 PM »

Offline nickagneta

  • James Naismith
  • *********************************
  • Posts: 48120
  • Tommy Points: 8794
  • President of Jaylen Brown Fan Club
Want to see the math.

Re: Introducing the J Rating (or J Points)
« Reply #3 on: March 01, 2019, 05:56:35 PM »

Offline RodyTur10

  • Jim Loscutoff
  • **
  • Posts: 2757
  • Tommy Points: 291
  • Always offline from 9pm till 3am
Yes, we like math. Please show, such that we can copy this rating system from you and sell it  ;D.

Like any rating system based on box score statistics your system probably doesn't take defense (enough) into account.
And does passing accuracy and rebound rate play a role in J-rating?

Re: Introducing the J Rating (or J Points)
« Reply #4 on: March 01, 2019, 06:17:53 PM »

Offline GreenEnvy

  • Antoine Walker
  • ****
  • Posts: 4551
  • Tommy Points: 1031
Without the formula this is completely useless.

Edit it in so we can actually discuss your rating system.
CELTICS 2024

Re: Introducing the J Rating (or J Points)
« Reply #5 on: April 15, 2019, 08:42:51 AM »

Offline gift

  • NCE
  • Ray Allen
  • ***
  • Posts: 3983
  • Tommy Points: 291
I finished putting together the J Ratings for the entire season. Here are the Celtics ratings:



For comparison, here are the top 20 for the league:


Re: Introducing the J Rating (or J Points)
« Reply #6 on: April 15, 2019, 10:25:06 AM »

Offline Moranis

  • James Naismith
  • *********************************
  • Posts: 33461
  • Tommy Points: 1533
Obviously without knowing the formula and what is accounted for and how, it makes it more difficult to offer critique, however, looking at the rankings they appear to about what you would expect if you were putting something like this together.  Harden and Giannis at 1 and 2 and a good deal better than the next group, who have Embiid, George, Curry, Davis, and Durant all pretty closely grouped, followed by Lebron, Westbrook, and Kawhi.  I mean looking at the stats for the season, that seems about right.  I'd definitely be interested to see how much games played alters things.
2023 Historical Draft - Brooklyn Nets - 9th pick

Bigs - Pau, Amar'e, Issel, McGinnis, Roundfield
Wings - Dantley, Bowen, J. Jackson
Guards - Cheeks, Petrovic, Buse, Rip

Re: Introducing the J Rating (or J Points)
« Reply #7 on: April 15, 2019, 11:13:19 AM »

Offline footey

  • Reggie Lewis
  • ***************
  • Posts: 15965
  • Tommy Points: 1833
Is that you, Bernie (Madoff)?


Re: Introducing the J Rating (or J Points)
« Reply #8 on: April 15, 2019, 11:56:32 AM »

Offline Big333223

  • NCE
  • Tiny Archibald
  • *******
  • Posts: 7489
  • Tommy Points: 741
I'm also interested in the methodology.
1957, 1959, 1960, 1961, 1962, 1963, 1964, 1965, 1966, 1968, 1969, 1974, 1976, 1981, 1984, 1986, 2008

Re: Introducing the J Rating (or J Points)
« Reply #9 on: April 15, 2019, 12:06:36 PM »

Offline Rosco917

  • Paul Silas
  • ******
  • Posts: 6108
  • Tommy Points: 559
Sounds interesting, but I need a hint on how these numbers are formulated.

Re: Introducing the J Rating (or J Points)
« Reply #10 on: April 15, 2019, 12:26:16 PM »

Offline Sophomore

  • Paul Silas
  • ******
  • Posts: 6227
  • Tommy Points: 823
Interested to see Baynes ranked so low. He's clearly critical to the team's success.

Re: Introducing the J Rating (or J Points)
« Reply #11 on: April 15, 2019, 12:50:05 PM »

Offline gift

  • NCE
  • Ray Allen
  • ***
  • Posts: 3983
  • Tommy Points: 291
Obviously without knowing the formula and what is accounted for and how, it makes it more difficult to offer critique, however, looking at the rankings they appear to about what you would expect if you were putting something like this together.  Harden and Giannis at 1 and 2 and a good deal better than the next group, who have Embiid, George, Curry, Davis, and Durant all pretty closely grouped, followed by Lebron, Westbrook, and Kawhi.  I mean looking at the stats for the season, that seems about right.  I'd definitely be interested to see how much games played alters things.

The rating is just an average of each game's rating. So games played doesn't affect the rating. You'd have to take into consideration separately the value of a 1.07 over 55 games (Lebron) vs. a 1.06 over 73 games (Westbrook).

Re: Introducing the J Rating (or J Points)
« Reply #12 on: April 15, 2019, 12:53:15 PM »

Offline gift

  • NCE
  • Ray Allen
  • ***
  • Posts: 3983
  • Tommy Points: 291
Interested to see Baynes ranked so low. He's clearly critical to the team's success.

I think a knock on Baynes in terms of his rating here is his only 16.1 minutes per game average. Limits his impact somewhat.

Re: Introducing the J Rating (or J Points)
« Reply #13 on: April 15, 2019, 01:00:28 PM »

Offline footey

  • Reggie Lewis
  • ***************
  • Posts: 15965
  • Tommy Points: 1833
Interested to see Baynes ranked so low. He's clearly critical to the team's success.

I think a knock on Baynes in terms of his rating here is his only 16.1 minutes per game average. Limits his impact somewhat.

So everything being equal, the more minutes you play per game the higher your J Rating?  What is the correlation?

Re: Introducing the J Rating (or J Points)
« Reply #14 on: April 15, 2019, 01:00:32 PM »

Offline gift

  • NCE
  • Ray Allen
  • ***
  • Posts: 3983
  • Tommy Points: 291
Sounds interesting, but I need a hint on how these numbers are formulated.

I won't give away the exact formula because my methodology is somewhat outdated (I devised drafts of it maybe 15 years ago, but took me years to get to a point where I could access the data I needed in an efficient way). Also, part of my interest in actually playing this out is judging my ratings vs. general perceptions of nba players' performances.

However, I'm getting everything from traditional box scores and weighing TS% and +/-. I know how flawed this can be, but that's why I say this rating does not tell you how good a player is, but how good their performance was.

I'm working on a new rating system, but I think I'm going to compile seasons 2001-present first to help me see particular areas for improvement.