How to make players play like themselves - The ATD Guide

[CelticLG]

Hi guys. Many users have asked me for advice on how I've edited the tendencies in my various rosters in the attempt of getting players to play the same way they would in real life. I preferred to create a discussion in order to be public, and to be able to confront each other, and improve the game together.

I've decided to make this little thread, for those interested, which will contain a series of extracts from my "Tendencies guide" that will be released with my roster shortly: https://forums.operationsports.com/f...il-roster.html

As you know, the default 2K roster has huge problems with tendecies. Many times they are random values (like high stepback or eurostep numbers assigned to big men who have never tried such a movement in their entire life), many times they are dated (like Blake Griffin's alley-oop and dunk tendencies) and other times they're only superficial values (like the 0 in stepback tendency for Doncic at the beginning of the year, or the current 0 in spin layup tendency gave to Siakam) and much more.

Spacing, shot selection, drive and kick, transition and all the general game flow seem very far from the game of basketball we see every day.

Luckly, at the same time 2K gives us all (or almost all) the tools to fix the problems.

In this thread I will try to talk, with a message every 1/2 days, about how to improve the tendencies in order to create a game experience as close to reality as possible.

Let's begin

[CelticLG]

Re: How to make players play like themselves - The ATD Guide

Part 1: Dish to open man tendency


I'd like to start with this one because in my opinion it's one of the most important tendencies to edit if you want to have better gameplay.

With the default roster the ball movement is pretty slow and static, you see very few drive-and-kick (this year particularly) and it's very difficult to see "the extra-pass".

The main problem is the assignment of low dish values to the main ball-handlers and passers of the teams.

We'll take the Sixers and their dish to open man tendency as an example.

(Players in bold are ball-handlers)

Burks Alec (20)
Embiid Joel (20)
Harris Tobias (30)
Horford Al (45)
Korkmaz Furkan (50)
Milton Shake (70)
Neto Raul (60)
O'Quinn Kyle (30)
Pelle Norvel (73)
Richardson Josh (25)
Robinson III Glenn (15)
Scott Mike (40)
Shayok Marial (45)
Simmons Ben (50)
Smith Zhaire (45)
Thybulle Matisse (30)

Ball handlers have an average value of 45 dish to open man tendency, team average is 40.

Sixers are one of the best teams in assists per game this season, so we need to raise this value.

You can use many reference statistics, such as average passes, assists, secondary assists and more. The value that I absolutely prefer and I think more precise is the potential assist, which does not depend so much on the quality of teammates as the actual assist.

Another statistic we'll need are the minutes played (I usually use all players with at least 200 minutes played the season, but in this particular case I have included all the players), so that we can calculate the potential assists divided for each minute.

Once the value is calculated, we can multiply it by 250, and round it up to the next multiple of 5, obtaining the new tendencies.

Now we can compare the new and old values of the Sixers' players (ball-handlers are marked in yellow)

Conclusions:

- average tendency is higher
- average ball handler tendency is way higher (+42%), this will lead to more passes and drive-and-kicks, improving overall ball movement
- Simmons tedency is 100, among players with at least 1500 minutes he's 5th in the league for potential assists per minute, behind only LeBron, Rubio, Luka and Trae. At the same time we will have to balance his drive, attack strong and driving layup/dunk tendencies, in order to have variety between passes and shots during drive to the basket situations.
- Burks is handling the ball more often than in the past two seasons
- shooters like Korkmaz normally have a low dish tendency (not 50 for sure) because... they just shot
- Norvel Pelle is a big man more similar to Hassan Whiteside than Nikola Jokic, how could you give him 73 dish tendency?

Hoping it might have helped, let me know what you think.

[Permatron]

Re: How to make players play like themselves - The ATD Guide

Great breakdown! Appreciate the work.


Sent from my iPhone using Tapatalk

[Vedreanu]

Re: How to make players play like themselves - The ATD Guide

I really like this idea, please keep the post flowing!!

[Vedreanu]

Re: How to make players play like themselves - The ATD Guide

Quick question:

Have you edited Miami and Milwaukee rosters yet?
Because, I've let them play (CPU vs CPU) and so far I've seen very little three pointers.

Also, Robinson hasnt shot once at the basket in 11 minutes he played so far. Its strange, eventhough I can see he has his 3p tends set pretty high.


These are the stats from the half time :
(off to sleep, the game is streaming on YT )

[Real2KInsider]

Re: How to make players play like themselves - The ATD Guide

Part 1: Dish to open man tendency

I would like to clarify something. This tendency does NOT work independently of itself.
It works hand in hand WITH the Attack Strong (Finish) tendency.

50 Attack / 50 Dish
75 Attack / 75 Dish

These players dish rates would function exactly the same from a drive & kick perspective.
The 75 Dish player would not pass more on drives.

Likewise...

50 Attack / 20 Dish
80 Attack / 50 Dish

These players would also pass out at similar rates. The game cares about the 30-point gap, not the total value.

It is unintuitive, but that's how it's programmed.

We'll take the Sixers and their dish to open man tendency as an example.

(Players in bold are ball-handlers)

Burks Alec (20)
Embiid Joel (20)
Harris Tobias (30)
Horford Al (45)
Korkmaz Furkan (50)
Milton Shake (70)
Neto Raul (60)
O'Quinn Kyle (30)
Pelle Norvel (73)
Richardson Josh (25)
Robinson III Glenn (15)
Scott Mike (40)
Shayok Marial (45)
Simmons Ben (50)
Smith Zhaire (45)
Thybulle Matisse (30)

As such, the posted data is incomplete / inaccurate.

Attack / Dish (Net)

POINT GUARDS
Simmons: 50/50 (0)
Neto: 15/60 (+45)

SHOOTING GUARDS
Richardson: 45/25 (-20)
Burks: 50/20 (-30)
Thybulle: 65/30 (-35)
Milton: 30/70 (+40)
Smith: 15/45 (+30)

SMALL FORWARDS
Harris: 65/35 (-30)
Korkmaz: 25/50 (+25)
Robinson: 35/15 (-20)

POWER FORWARDS
Horford: 65/45 (-20)
Scott: 60/40 (-20)

CENTERS
Embiid: 75/20 (-55)
O'Quinn: 65/30 (-35)
Pelle: 30/73 (+43)

Like you, I'm in favor of raising Dish across the board for the vast majority of players.
However, the solution is not as cut & dry as it would otherwise seem.
In some cases editing a player's Attack tendency (or both) would be the correct course of action.

shooters like Korkmaz normally have a low dish tendency (not 50 for sure) because... they just shot

This is an example where it might be correct to increase his Attack tendency.

- Norvel Pelle is a big man more similar to Hassan Whiteside than Nikola Jokic, how could you give him 73 dish tendency?

I just want to clarify that every G-Leaguer/two-way (and even drafted players) have their tendencies copy/pasted from an existing NBA player or template. At the rate these players get added to the game nowadays it would be impossible to keep up otherwise. No player has their tendencies done from scratch.

As an example, Kyle Alexander, Ray Spalding, Alen Smailagic all have the same tendencies.

Speed is more important than accuracy in this regard. These players get added to the game the moment they make their debuts. The vast majority of the time, these players never even take the court in-game. It only becomes an issue if any of them take a major leap in ability (at which point they'd need a revision due to a reformed playstyle anyway).

I do not know the exact process 2K goes through when they add a player to the game but they definitely don't consult whomever the Tendency specialist is.

[CelticLG]

Re: How to make players play like themselves - The ATD Guide

Man, really appreciated all the breakdowns, but can you group all messages into 1/2/3 (maybe with spoilers, etc)? If we can avoid 10 messages in a row I think it's better, in order to keep the thread clean. Thanks

[Vedreanu]

Re: How to make players play like themselves - The ATD Guide

Hey, CelticLG, how do you comment (3p) stats from the above game with your roster?

I am using 3p boosted sliders.

[CelticLG]

Re: How to make players play like themselves - The ATD Guide

Hey, CelticLG, how do you comment (3p) stats from the above game with your roster?

I am using 3p boosted sliders.

You mean % or shot distribution?

[Vedreanu]

Re: How to make players play like themselves - The ATD Guide

You mean % or shot distribution?

Not the percentage. The amount of 3pts taken. Pretty low, lower then with generic roster.

After that match I put coaches' style 1 for both Miami and Milwaukee to "SHOOTING THREE". Also did that for assistant coaches.
And turned POEs to "Shot at will".

Only slightly more threes were shot, nowhere near the realistic amount.

The problem might be in what this user has stated, also, maybe
Drive contact shot frequency
slider is making problems.

I lowered it from 100.

What sliders do you use with this roster of yours (ATD NEW)?

[CelticLG]

Re: How to make players play like themselves - The ATD Guide

Not the percentage. The amount of 3pts taken. Pretty low, lower then with generic roster.



After that match I put coaches' style 1 for both Miami and Milwaukee to "SHOOTING THREE". Also did that for assistant coaches.

And turned POEs to "Shot at will".



Only slightly more threes were shot, nowhere near the realistic amount.



The problem might be in what this user has stated, also, maybe

Drive contact shot frequency

slider is making problems.



I lowered it from 100.



What sliders do you use with this roster of yours (ATD NEW)?

My sliders are stil a WIP, but a lo my of thing affects 3pt shooting. Zone usage sliders, play-types, playbook plays, gameplay sliders, a loro of tendencies (pullup 3pt, drive, post up, etc)

[Vedreanu]

Re: How to make players play like themselves - The ATD Guide

You have written before about making a "random generator" with your settings/sliders. I am very interested to hear whatever you have to say on that topic.

How is that generator working, are you changing some settings manually or is it more sophisticated?


I am asking because when I watch CPU vs CPU sims, there seem to be to little turnarounds of big leads. Also, what bothers me the most - the same players will always have almost the same shot statistic with very very low variance.

For instance, Butler would virtually always shoot around 40% with 20ish shots...
Giannis is almost always above 50%... etc...
There is no "unknown" factor in players behaviour. Maybe that could be bypassed somehow?


Do you have an idea how to deal with that?
I tried several options, mostly with badges, but no luck...

[CelticLG]

Re: How to make players play like themselves - The ATD Guide

Part 2: Post up tendency

As many of you surely know the post up is a much less used play in the modern NBA than in the past due to its lack of efficiency. The Nets only use 0.4 post ups per game this season.

The post up has experienced a slow decline, until this year, where for the first time in history the average post-up for each team is less than 5.

The idea is simple: less post ups, more ball movement, more pace, more triples.

So we must try to reduce the post up tendencies of some players, which are exaggerated compared to how they behave in real life, especially this year.

We'll take the Bulls as an example, one of the least post up oriented teams in the whole league. These are their post up tendencies (big men in bold):

Arcidiacono Ryan (0)
Carter Jr. Wendell (55)
Dunn Kris (10)
Felicio Cristiano (35)
Gafford Daniel (5)
Harrison Shaquille (0)
Hutchison Chandler (0)
Kornet Luke (20)
LaVine Zach (10)
Markkanen Lauri (30)
Mokoka Adam (0)
Porter Jr. Otto (10)
Satoransky Tomas (0)
Strus Max (0)
Valentine Denzel (0)
White Coby (0)
Young Thaddeus (45)

Average post up tendency is 12.94 (but we have a lot of 0s), and the average bigs post up tendency is 31.67, too high for us standards.

First of all we need to find the stats from nba.com (here), in this case post ups and minutes played, in order to calculate the post up per minute (we can once again use the criteria "at least 200 minutes played"):

Once this value is obtained we can multiply it by 250 (same number as the dish to open man tendency), and round it by the number closest to 5.

As you can see Embiid will still maintain a high tendency, same as Davis or Aldridge, while players who use the post up less frequently will be decreased.

Let's see the results with our Bulls (bigs are marked in yellow):

So we've globally reduce the post up tnndecy, that was too high for a lot of players.

Felicio only used 2 post ups in almost 400 minutes this year, LaVine has only attempted 9 shots from the post in the whole season. Young and Carter Jr. are used more as pick and roll/pop bigs than post up players, same for Kornet that shoots a lot of threes but never post up.

It's an important edit that makes the game smoother and less static, many teams will benefit a lot from it, while others will only need a few small changes (for example the Hawks below).

See you next post!

[Real2KInsider]

Re: How to make players play like themselves - The ATD Guide

We'll take the Bulls as an example, one of the least post up oriented teams in the whole league. These are their post up tendencies.

It should be noted that these figures were based on 2019 production. A direct comparison to 2020 production is going to lead to false conclusions.

Felicio only used 2 post ups in almost 400 minutes this year, LaVine has only attempted 9 shots from the post in the whole season. Young and Carter Jr. are used more as pick and roll/pop bigs than post up players, same for Kornet that shoots a lot of threes but never post up.

Kornet DID post up last season, strange as that may sound.
Young has always been a post player rather than a pick & pop player (until this season).
LaVine has 20 Postups so I assume this is a typo.

2019 Postups per 36 mins (Total Postups)
5.9 Young (414) - Pacers - 10th in total possessions
4.4 Carter (134)
2.8 Markkanen (130)
1.8 Kornet (39) - Knicks
1.2 Felicio (25)
0.9 Porter (42)
0.6 Dunn (23)
0.5 LaVine (28)
0.3 Satoransky (18)
0.2 Hutchison (6)
0.1 Harrison (4)
0.0 Valentine (3) - 2018 stats
0.0 Arcidiacono (1)

Other 2019 Bulls
6.6 Robin Lopez (293) - Made up 32% of his possessions, 2nd most in NBA
5.1 Bobby Portis (182)
3.5 Jabari Parker (169)

Worth noting that Jim Boylan was installed mid-season and coaches typically need training camp / preseason to instill their vision.

2020 Postups per 36 mins (Total Post)
2.2 (-2.2) Carter (78)
2.1 (-3.4) Young (94)
1.2 (-1.6) Markkanen (50)
0.8 (NEW) Gafford (15)
0.3 (-0.2) LaVine (20)
0.3 (NC) Satoransky (17)
0.3 (-1.5) Kornet (6)
0.3 (-0.3) Porter (3)
0.1 (-0.5) Dunn (6)
0.1 (-1.1) Felicio (2)
0.0 (NEW) White (3)
0.0 (-0.2) Hutchison (1)
0.0 (-0.1) Harrison (1)
0.0 (NC) Valentine (0)
0.0 (NC) Arcidiacono (1)

It's an important edit that makes the game smoother and less static, many teams will benefit a lot from it, while others will only need a few small changes

The way this post was initially framed, it's correcting a league-wise issue. Yet in this example the team had a drastic change in philosophy.

Based on your graph, an average NBA team only lost 1.0 Post possessions per 36 minutes. The Bulls very clearly reduced far more than this.

The league-wide decrease may very well be placebo as 2019 leaders were beset by injuries.
Blake Griffin (5th), Marc Gasol (7th), Kevin Durant (8th) combined for 1600 postups last year.
This year they have 140.

[CelticLG]

Re: How to make players play like themselves - The ATD Guide

It should be noted that these figures were based on 2019 production. A direct comparison to 2020 production is going to lead to false conclusions.

Sure, we're just updating the tendencies to actual values, and the trends they're following IRL