How do you turn stats into ratings?

[Real2KInsider]

Quote: Originally Posted by Sonicmage

This was the toughest problem to solve in generating shooting ratings based on real data. What do you do when a player shoots 3 for 4 from 3-point range, but last time he was in the league was 3 years ago when he shot 25 for 100? Do you trust his career average or something closer to his current average? How do you know he hasn't improved his 3-point shot significantly in his three year absence? I fudged the numbers for this example but it's not that different from a real situation.

The best example of such a player would be Sonny Weems, who has non-applicable Euro data over the course of his absence (he would require a hands-on approach regardless).

[Real2KInsider]

Quote: Originally Posted by iTofu

Then I did mid range and found that there was no clear formula no matter what range of feet I took.

2K has a formula for Close and Mid (as of 2K15, at least), though like their other formulas they sporadically follow it (and often ignore it for volume scorers like Kobe/Melo).

Part of the problem is 2K will edit a player mid-season after a hot/cold streak and then not follow up in the off-season when everything balances out. Or a player will get hurt and they won't touch him until he gets healthy (in the case of Kobe, ignoring how awful his performance was prior to that point)

[Real2KInsider]

NEW YORK KNICKS
2K Standing/Moving Midrange
90/88 Arron Afflalo (80-180 FG; 44.4%, 66% Ast'd)
85/78 Sasha Vujacic (104-141 FG; 43.2%, 69% Ast'd) - 2011 Stats
84/84 Carmelo Anthony (102-220 FG; 44.5%, 49% Ast'd)
84/76 Langston Galloway (59-136 FG; 43.4%, 51% Ast'd)
82/68 Kevin Seraphin (27-64 FG; 42.2%, 85% Ast'd)
80/78 Kristaps Porzingis (N/A)
77/72 Robin Lopez (13-33 FG; 39.4%, 100% Ast'd)
75/75 Jerian Grant (N/A)
75/72 Jose Calderon (51-131 FG; 38.9%, 59% Ast'd)
74/70 Kyle O'Quinn (20-56 FG; 35.7%, 65% Ast'd)
73/64 Derrick Williams (23-65 FG; 35.4%, 57% Ast'd)
70/68 Lance Thomas (50-149 FG; 33.6%, 84% Ast'd)
65/62 Cleanthony Early (7-35 FG; 20.0%, 71% Ast'd)
62/44 Lou Amundson (7-14 FG; 50.0%, 100% Ast'd)

In 2K15 when they added the "Moving" shot rating it seemed like they really screwed some of these up, with a lot of players having a higher moving shot than a standing shot rating (which fundamentally doesn't make sense).

It seems like that was corrected in 2K16, as every Moving here is equal to or less than the Standing shot, which jives with the way I approached the rating. Use the FG% as a Standing base, then adjust Moving based on assist%.

You can kinda sorta see them scale properly, though not cleanly. Basically everyone who shot 40% is in the 80s, including the low volume Seraphin (though excluding Amundson). Some of the "Moving" shots make no sense (compare Robin Lopez to Jose Calderon for a laugh).

Rashidi Ratings (roughly)
91/87 Carmelo Anthony (+7/+3) - Odd that 2K wouldn't stack him here per usual
90/82 Arron Afflalo (NC/-6)
87/83 Langston Galloway (+3/+7)
87/77 Sasha Vujacic (+2/-1) - IF we ignore his DREADFUL last 4 years in Europe.
80/72 Kristaps Porzingis (NC/-6)
79/69 Kevin Seraphin (-3/+1)
78/72 Jose Calderon (+3/NC) - Could bounce back if healthy. Shot 48% (39% Ast'd) in 2013.
75/75 Jerian Grant (N/A)
74/64 Robin Lopez (-3/-8) - He will probably take a ton more this year. 91-218 (41.7%) in his career.
74/64 Lou Amundson (+12/+20) - Fun fact: He is actually 26-58 (44.8%) in his career.
71/63 Kyle O'Quinn (-3/-7) - Should bounce back, shot 42% in his first two seasons
71/65 Derrick Williams (-2/+1)
68/58 Lance Thomas (-2/-10)
60/52 Cleanthony Early (-5/-10)

Bear in mind I haven't factored in any "HoF rating scale", whatever minimal effect that entails. It does not appear as if these were re-scaled like Speed/Rebounds/etc given how minor my changes are from the actual rating.

[Coachjosh21]

Good discussion guys. I started making note of each rating and tendency and their average number by position. At least it's a foundation to work from when creating historic players to fit in with players already on the official 2k roster.

In the end it seems to come down to testing in gameplay. A couple of you made a great point about how gameplay stats don't match sim stats 100%. How the game plays should come first. Then gameplay stats second. And sim stats are the lowest priority. Although I would think sim stats would be where you start when assigning ratings and tendencies to players.

[Real2KInsider]

Rating changes from 2K15 listed.

Carmelo Anthony
88 Overall (+2)
OFFENSE
86 Standing Layup
83 Driving Layup (+1)
87 Post Fade
64 Post Hook
77 Draw Foul (NEW)
84 Standing Close (+2)
84 Moving Close (+4)
84 Standing Mid (+3)
84 Moving Mid (+6)
78 Standing 3pt (+5)
77 Moving 3pt (+10)
79 Free Throw
88 Ball Control (-3)
66 Pass Vision (+7)
70 Pass IQ
68 Pass Accuracy (+2)
53 Offensive Rebound (-2)
55 Standing Dunk
75 Driving Dunk
50 Contact Dunk (NEW)
97 Shot IQ
98 Hands (+3)

Shooting scales did not change. Mid/3pt were reduced last year as a workaround for patch four, so Melo saw minimal change here.
Everything else doesn't seem to have a new scale and the minimal changes are related to season production (stats).

DEFENSE
66 Defensive Rebound (+6)
50 Block (-6)
74 Shot Contest
61 Steal (+2)
76 On-Ball Defense
70 Low Post Defense
78 Reaction Time (NEW)

Scales might have been tweaked leading to the minor changes (which will become clearer the deeper I get into this).

ATHLETICISM
72 Boxout (+10)
68 Lateral Quickness (-18)
82 Speed (-8)
80 Acceleration (-5)
74 Vertical
82 Strength (+2)
98 Stamina (+3)
82 Hustle (+5)
84 Durability (-1)

Lateral Quickness was reduced league-wide. Czar and OG did this to create the improved half-court gameplay.

Position Average
PG: 76 (last year this would have been in the 80s)
SG: 72
SF: 74 (wait what)
PF: 56 (meanwhile most PFs are SFs nowadays)
C: 45 (can't stand how 2K thinks modern PF/C are lumbering bigs circa 1985)

Melo's Speed/Acceleration (Quickness) are actually equal to what I've been giving him for years via his Draft Combine numbers (adjusting for age), but these are decreased across the board as well.

Stamina has a new scale, something Czar was working on to reflect changes in how fatigue accumulates (Usage matters).

INTANGIBLES
76 Pass Perception (+12)
60 Defensive Consistency (+10)
75 Pick & Roll Defense
74 Help Defense (+4)
98 Offensive Consistency (-1) - Because Hall of Fame system.
99 Emotion (+24)
88 Potential (-2)

Nothing really changed here beyond a desire to give Melo better defensive ratings.

With regard to Pass Perception the vast majority of players did not receive an update. In fact, only two other Knicks saw a change here: Lance Thomas 70 (+12) and Derrick Williams 62 (+9)

So really, the only scales that need to be evaluated are rebounding, steals, blocks, and maybe Passing Vision to a lesser extent (along with the athletic ratings).

[Real2KInsider]

With regard to the following ratings:
Standing/Moving Close
Pass Accuracy
Off/Def Rebound

2K's ratings here all match my own.

As I observed earlier in the month w/ Kobe, it's the first time they've actually followed their own rebound scales.

The ratings they did adjust probably don't need much tinkering but the ones they left alone (like Melo's dunk rating) are so bad.

FWIW applying my own changes I have Melo at 86 OVR. A worse finisher/post player, a better shooter, and a lesser defender.

[Coachjosh21]

Rashidi, have you shared your formulas somewhere? Simple stats such as shooting, assists, blocks, rebounds etc. would be easy to calculate ratings once you know what numbers to plug in

[Coachjosh21]

Ok guys I figured out the formulas for standing shots. It didn't take all that long to get it. Is there a reason that all the formulas aren't discovered and stickied to the top of this forum for all roster editors? Do 2K or roster editors stand to lose something if this gets released? I don't want to screw anyone over, but it seems like if I could figure out, so could others