ATD Sheitan Off-Season - Legendary CPU realism in-game & Trades & FA

[Sheitan]

Hey 2K community ! For those who don't know me, I am French, NBA fan for more than a decade now. I am working on NBA 2K for a year now in order to improve the realism in-game, mainly CPU behavior, and also the simulation side in MyNBA based on the ATD Roster from CelticLG.

After hundred of hours of testing, as from now, I get everything in NBA 2K. From attributes, to tendencies (simulation & in-game behavior), in-game sliders, coaching sliders and MyNBA sliders.

Please find here the project I started earlier this year

I am coming now with an Off-Season scenario based on the latest update of the ATD Roster with amazing improvments.
Please note that this scenario has been built to be mainly a CPU vs CPU.

PS5 scenario : Sheitan OffSeason (PSN: Sheytanlevrai)

List of changes :

- Correcting the broken attribute "speed with ball" - no more blow by

Spoiler

* If you were not on the ATD Discord you probably don't know but the attribute "speed with ball" is basically responsible for the "blow by" in half court attack where defenders cannot follow the ball handler with too high "speed with ball" value (basically above 70). Also the player size impacts a lot the "speed with ball", where basically a 6"11 like Giannis with 60 speed w/b is moving as fast as a 6"3 with 70.
* I compensated by adding the badge "downhill" HOF to every player so in transition players can still make the difference.

- Adjusting ratings/tendencies/badges.

Spoiler

* Tendencies with 99 values are broken such as Alleyoop. I lowered the max to 90 otherwise players with 99 do alleyoop too often in game (between 5 or 10 alleyoop/game). In the future we need a better distribution of those figures and not like it is currently with a group of players around 90-99, then around 50, then around 0. Because a player with a 50 alleyoop tendency won't basically do an alleyoop per game.
* Nerf defensive players such as Bledsoe and overrated players like Kemba.

- Adjust CPU sliders. no more blow by and way better pick&roll defense

Spoiler

* Amazing realism in-game with turnovers, faults, good balance between contested/uncontested going in. Good equilibrium between the efficiency of layup, close shot, mid and threes. Good balance as well with contact in the paint.
* defensive consistency is the key because it defines the probability that the defender will freeze. By increasing this slider the defense react better even if they are bad defenders. It decreases the number of blow by and lead to better offensive IQ

- Correct season awards & correct draft lottery

Spoiler

* Jokic MVP, Bucks winning the NBA title with Giannias final MVP and correct order during the draft.

- Amazing Draft class with Archetypes & Intangibles leading to an incredible realism !A CPU team with a bad scoot can now choose a bad player with a top 10 pick

Spoiler

* I defined several Archetypes (players DNA) for attributes & tendencies that I use as a base for each player of the draft (offensive, defensive, two-way, athletic, playmaker, all-around, three point, mid range, post for 3 positions : guards, forward & center). Then I import the attributes and tendencies of the appropriate archetype on each players. Last but not least, I might have done a global +1 for future superstar or global -2 or -3 or even more for future role players (late 1rst round picks).
* The goal of Archetypes is to have future players behaving like current ones, in terms of ratings and in terms of tendencies. So in your scenario in year+2, +3 or +5 you will have players acting accordingly to their archetype in simulation stats and also in-game.
* I played with "Intangibles" so that every player has an overall rating around 75. It leads to some bad choices by the CPU if their Head Scoot is not able to correctly see the intangible and the potential. Which is an amazing improvement in term of realism.
* Example : we have around 20-25 players with a 75 overall, some of them have great potential (above 85), some of them have ****y potential (between 75 and 80). A CPU team with a bad scoot can now choose a bad player with a top 10 pick !

- Nerf Progression & Training

Spoiler

* The in-season training feature is basically broken, so I nerfed it with a 1 value to avoid players having a global +5 in attributes during the season.
* Attributes progression are also really bad from every roster creator unfortunately, with so high values that lead to some +5 in each attribute each season, leading to a bunch of +85 overall players where everyone is a Stephen Curry with 90-95 in 3points after few years.

Therefore after some tests, I decided to fill the broken attributes such as "Close, mid-range, three-point" with a 1 value for "growth phase" and even with those values players with a great potential such as Doncic can have a +3 per year. But at least it is limited to amazing talents.
The others attributes can have a max 50 value, and "driving layup" is a +25 otherwise every player will get a +90 in layup after few years at it brokes the game (simulation & in-game).

- Trades & Contract Sliders - Awesome Free Agency with contenders that can sign good players with their mid-level exception or with a minimum contract !

Spoiler

* There's already a discount in forward draft picks so even with a figure of 100 a draft Y+3 has a lower value than a +2 and so on. And selling and rebuilding teams like forward draft picks so there's a need to increase their value.
* Player type discounts all to 100 to maximize the value of each players and so CPU is doing better in term of choices, trying to trade for the better player where they face weaknesses.
* I kept the original Contract sliders because they are amazing with the roster.
IMPORTANT :
Contract sliders define the "expect salary" for a player.
"Contract Negotiation Difficulty" defines the importance of the salary in the free-agent choice.
"Importance of non financial parameter" needs to be at 100.

You can see it in the free-agency when a player receives multiple offers from teams, with high values in "Contract Negotiation Difficulty" he won't consider offers with a lower salary than the expected salary. Whereas with really low "Contract Negotiation Difficulty" the player might consider to accept an offer with a lower salary than the expected salary, in regards to its characteristics (importance of winning, importance of salary, ...). Therefore a "Contract Negotiation Difficulty" with a value of 0 can lead to good players around 78 to 80 overall to sign for a contender for a mid-level exception or a minimum salary.

In my scenario I have seen the Lakers signing Zeller (79 overall) with a minimum salary, and some players like Gordon & Bullock (3&D with 78 overall) for minimum as well.

There is no impact on the User free-agency. Unfortunately the user cannot benefit from the low value of "Contract Negotiation Difficulty" because if you offer a bad salary to a pretty good player the pop-up "the player X refused the offer" will appear, even if the global offer has a 70 value. (consideration value from the free agent required so that they can sign).

It's not bad as it provides a great challenge for you guys !

Also the original contract sliders are way better than the one defined by roster creators, for the reason that some average players can sign crazy contracts as it happens in real life (cf : Hayward, Kemba Walker, John Wall, Schroder, Dragic, ...).

With the original sliders a player around 80 will ask for 15m$, and a player above 83 will ask for the max which is legit and corresponds to the reality.

- Draft 2022 ready !

Spoiler

* ID : Sheitan 2022 (PSN : Sheytanlevrai)
* As mentioned above, I used the online draft "2022 Draft(50 real prospects)" where I imported my Archetype DNA for 18 players and I lowered their overall rating with the Intangibles. So that CPU with bad scoot will be able to choice bad players in the top10 in the next year draft.

Future improvements and tips :

To be updated :

- Coaches & coaching tendency with "run play frequency" with a maximum value of 50.

Spoiler

This change will speed up the tempo in-game where players will have more freedom and if they see an opportunity in the defense they will take it.

MyNBA - offseason - TO DO :

Spoiler

- After the draft : turn the Intangibles to 100 so that drafted players can be starters. Because CPU builds his team based on the global overall value.
- Free Agency : In case good old players with high rating (around 80) are not selected on the day10 go to "contract sliders" and turn down to 0 the following values (durability, minutes, player fame). Those players will decrease their "expected salary" and might sign at a really low cost especially old players.

Then turn it back to their original values ((durability=40, minutes=60, player fame=20)

TO DO in-game for a CPU vs CPU game :

Spoiler

Take control of both teams and change the following :
- Offensive play to Custom instead of "pound the ball inside"
- Defensive play to automatic everywhere and "catch hedge" for on ball screen (center)
- Take a timeout with one team to make those changes effective

DO NOT increase the "help defense" to 100.
In fact it is even better if you lower it, with low "help defense" value the defense is acting like in a "zone". It provides a better pick&roll defense where the center goes under instead of switching badly and opening a free path to the rim.

I would say "help defense" should be floating between 25 and 75.
25 is basically a zone, where the defense is focus only in protecting the paint.75 there are many rotations. 50 is a good equilibrium. and 35 & 65 values give perfect behavior for both strategies.

Anyway you need to have the corrects players to play the correct defensive strategy. You can play switch with 75 help defense with players like Cousins or Whiteside.

An example of what my changes can bring in-game (CPU vs CPU) :

Spoiler

- Pick&Roll attack from Haliburton
- Zion goes under to defend the paint (due to the increase of defensive consistency he has a better reaction time and low help defense value tells him to go under)
- Haliburton sees it and his only option is a mid-range floater from the free throw line

https://www.youtube.com/watch?v=yo7bQpZ6DSc

Please share any comments or remarks guys. I work a lot so my free time is quite limited but I will try to answer to everyone.

[Sheitan]

As you have probably noticed if you play against the CPU, tendencies are the determinant factor in-game and in simulated games for a CPU player.

Where attributes are the characteristics of a player, badges are his skills, tendencies are the player IQ. It represents the ability that a player X will use his strength (attributes & badges) accordingly.

Therefore there is a need to define tendencies in a very accurate way when we create rosters, and also tendencies need to evolve as attributes and badges do.

Unfortunately 2K does not consider it as it should be. That's why I propose to create a "tendency training camp" in offseason. Through a spreadsheet such as the ATD event spreadsheet.

I give you some real-life examples :

- Since few years Lebron James is using the spin layup, he did not have it originally. In 2K this evolution can be represented by an evolution of tendencies. But it does not exist.

- Since few years, many players are copying James Harden and using more and more the stepback 3. As well this skill is represented in tendencies. A player can have the badge "circus three" but if he has 0 or below 50 in stepback3 tendency he won't use the move that often.

- Since few years, more and more players are also using the floater. Because defense tend to go under during the pick&roll to let an open mid-range to the offense.

- Also you can have a guy like Jeramy Grant who was a bench players at the Nuggets (not shooting often) and became Franchise Player at the Pistons, therefore his shot tendency increased a lot. And this evolution is missing in 2K.

Tendencies are left behind, that's why I created "Archetypes" in the Player DNA module, to define players ratings/tendencies based on it. (offensive, defensive, two-way, athletic, playmaker, all-around, three point, mid range, post for 3 positions : guards, forward & center).

Then I will create some "tendencies training camp" in the ATD event spreadsheet and in each offseason I propose that users update players tendencies accordingly to the spreadsheet. For instance it could be "stepback 3 training with James Harden", "floater training with Tony Parker", "fadeaway training with KD", "spin layup training with LBJ", "crossover training with AI" and so on...

There is clearly a lack of tendencies evolution in NBA 2K and we are gonna create it by ourselves.

[Chazza90]

Any updates?