All posts

Basketball Coach Game Update: Touch Levels and CPU Logic

, 8 min read

The November Basketball Coach game update brings several new features to the site. Most prominently, you can now set how much of the offense runs through each player using touch levels.

Touches in basketball are the opportunities each player gets to have the ball in a given game. They are decided directly or indirectly whether it is pickup or a professional setting. I tend to view basketball as a role-based game, where each player slots into a given archetype for their team. That leads to a hierarchy for each type of role: who takes the toughest matchup first? Who is the primary rim protector? And who are the ones we want with the ball most?

This post walks through how Basketball Coach answers that last question, then covers the other areas that changed:

  • The in-game Coach tab
  • Coaching plan controls
  • Computer coaching logic
  • League Mode quality-of-life

Touches: who the offense runs through

Touches are the lead feature that was shipped in this update.

The four levels

You set a level for each player in Your plan > Players > Share of the plays:

Each touch level is based on the real-life usage rate of a representative player who fills that role on offense for his team. In most Celtics lineups, you expect Hugo Gonzalez to be the fourth or fifth option. A player like Karl-Anthony Towns, on the other hand, is a near-guaranteed first or second option in any lineup he plays in:

levelleague positionusage it asks forthe real player nearest it
Minimal5th percentile11.8%Hugo Gonzalez
Less25th15.9%Duncan Robinson
Defaulthis ownhis real usageN/A
More85th25.9%Karl-Anthony Towns
Most99th34.1%Kawhi Leonard

The inspiration and full implementation of the touches system deserve their own deep-dive post, so I’ll keep this one brief. Here are the highlights of how it works today:

  • Levels are set per player, before the game or at a stoppage.
  • Five players share one ball. Two players on Most do not both get 34% usage. A touch level is a request for a usage range, not a promise.
  • “Touches” is the coaching word. What the game plans and counts is plays finished: a shot, free throws, or a turnover.
  • A player pushed above his usual share gets plays called for him in the play types he really runs. With Luke Kennard on Most, the off-screen plays called for him went from about 1 a game to about 5.

Plays Finished Panel

This new panel gives you a real-time view of how your planned distribution of touches is going for the players in the game:

  • The fill is his plays finished tonight. The tick is his plan for the game. The tag is his level.
  • The plan is one number for the game. It moves only when a coach changes a level, the featured player, or minutes.
  • A bar lights up as behind or ahead only when the gap is larger than ordinary game-to-game noise.
  • The panel is in every mode’s Coach tab.
  • Bars finish a little short of the tick on average due to several factors, including what actually happens during the course of the game.

The Feature in Action

As a fun experiment, I loaded up a Quick Game as the Knicks against the Nuggets. The only setting I changed was Mikal Bridges’ touch level, which I set to Most. Given his struggles to start the real season, maybe making him a primary option would help?

In the Plays finished panel, note how many plays Bridges is planned for on his default setting:

Now notice the jump in 12 more planned plays when his touch level is set to Most:

Unfortunately, we did not win the game, although Mikal had a decent individual performance:


The “Where it’s coming from” panel

This is the second headline feature. The Coach tab now shows the evidence a coach would act on, and lets you act on it:

  • It shows which screens, help defenders, and on-ball defenders are being beaten, counted in trips and points.
  • A staff call button queues the fix for the next stoppage.
  • It is in the Coach tab in every mode.

This feature supports one of the core tenets of the game: giving user coaches detailed, actionable feedback about everything happening in the game. I’m still tinkering with how often it speaks up and the thresholds behind it, but I believe it will help coaches of all levels spot problems and advantages. I’m happy to hear feedback on this one, as it is likely to change soon.


New plan controls

These dials are either new or now work more realistically.

Who brings the ball up

Your Plan –> Players –> Brings it up:

A direct dial for this became important to me once I saw a possession in which Mitchell Robinson brought the ball up the court. I’ll always have love for Mitch as a long-time Knick, but that’s unrealistic. It turned out that defensive rebounders bringing the ball up was over-tuned, especially for big men. That should be reserved for more realistic cases like Jokic, Bam, or Giannis. If you’d like to play around with the setting anyway, you can. Do so at your own risk.

Deeper Scheme Identity

New settings control how much of your offense goes to off-ball movement and to isolations. Your Plan –> Shape of a trip plays:

These settings let you dictate how your team plays on a given possession. Leaning either way has pros and cons, and I’ll keep examining the data on how much they help.

Action chains

Action chains are packages of pre-set steps you can pick for a play:

This is another experimental addition, meant to give you more creativity and a personal touch on your team’s offensive structure. How your team moves through its reads as a chain unfolds should be in your hands. A small word of advice: players can fail to execute an action, depending on their attributes and tendencies, so be wary of running a complicated offense with every roster.


Improved CPU Coaching Logic

The computer coach now uses the same plan and dials you do, and has more tools for adjusting in specific situations. Some highlights:

Cleanup:

  • The old Coach Mode “AI coach” logic is gone.

Rotations. CPU coaches now:

  • Weigh their depth against the opponent they are facing
  • Account for tired legs better
  • Avoid five-man lineups with no center, outside of specific scenarios
  • Aim to keep a real ball handler on the floor
  • On developing teams, protect their young players’ minutes

Playoffs. CPU coaches play more realistic playoff rotations, based on historical playoff data:

  • Star players play about 3 more minutes on average
  • The end of the bench loses minutes, or gets cut from the rotation entirely
  • The rotation does not loosen as a series goes on

Staff strength (League Mode). You can set the strength of opposing coaching staffs: Weaker, Real, Sharper, or Elite.

  • Real is based closely on the data I examined, but is admittedly a subjective view of how each coach runs his team.
  • Sharper and Elite staffs react sooner and make more changes. Weaker staffs react more slowly.

So far, my early measurements do not show these settings changing how often teams win. I’m continuing to work on coaching adjustments, and on the countless basketball interactions and coverages beneath them, to better reflect real life. That should eventually produce a clear difference between a bad coach and an elite one.


League Mode

Quality-of-life and realism changes here. Some highlights:

  • Your game is now saved at the final whistle. Closing the tab on the post-game screen no longer loses it, and a reload cannot replay a loss.
  • A new League maintains information on who is injured as of Game 1, and shows your injury list before the season starts:

Note: shortly after this update, it’s been announced that Porzingis is stepping away from basketball. A subsequent update will remove him from the Warriors’ active roster indefinitely.

  • Opening-night starters for all 30 teams.
  • Real heights for most players.
  • Scouting shows where to attack a real player, if they have noticeable weaknesses. A new Attack column in the Scouting tab names his weak spot, drawn from the same measures that decide negative traits:
  • The locker room no longer suggests lineup changes off a few minutes of data. Lineup cards wait for real samples and for the 10th game.
  • Simulating games in the postseason no longer freezes the page.

Sim Balancing Changes

These changes are aimed at the general basketball logic of the simulation. I may write a fuller post on any one of them when I find the time, as each has an interesting story:

  • The absence of a star-level creator (think Luka or Cade) has a more noticeable impact on his team’s offense, and his teammates shoot worse without him.
  • Offense in general is less effective without a ball handler on the court. Expect shots later in the possession, more contested jump shots, and fewer rim attempts and corner threes.
  • A good perimeter defender has a more noticeable impact on how often his man catches the ball and shoots.
  • A good rim protector turns more layups and dunks into floaters. The goal is to capture the deterrence of an intimidating interior defender like Wemby.
  • A defense in the penalty fouls less.
  • The screener in a pick and roll now reads the coverage and can slip a switch, depending on his offense IQ.
  • A player catching a pass reads the incoming closeout before he shoots, and his offense IQ shapes that read.

Fixes and small changes

  • Timeouts are stoppages for both teams in Quick Game and Playoffs.
  • A new game no longer offers to watch back the last game’s skip.
  • Shot-clock violations count as team turnovers.
  • Blocked browser storage can no longer blank the site. I may add more ways to save progress in the future, although browser storage is super convenient.
  • Coach Mode’s dead substitution-difficulty settings are retired.

Try It Out

If you want to test the new update, run the same experiment I did: pick a role player on your team, set his touch level to Most, and watch the Plays finished panel. If you did this, how did it go? I’d love to hear examples of it working out, or going very wrong. Reach out on Reddit, X, or the contact page on the site. Send box scores!

Next up is either a deeper post on how the touches system was built, or one explaining how attributes and traits work. Which would you rather read first? Let me know, and subscribe to the newsletter if you haven’t already, so you get notified directly once it’s out.

Thanks for playing, and for reading.