AI Depth Chart Matrix Management

[ukSUNSfan]

As I'm still new to the game (played 2021 since June), there might be a way of doing this that I'm not aware of so apologies if so.

The AI depth chart/matrix management doesn't seem to match the way it's typically done in real life in the sense that rarely do you see real life Head Coaches take all 5 starters off at the same time. In 2021 and in my quick first look at 2022, asking the AI t set the entire matrix often results in them having the starters minutes all aligned together, and likewise the bench players.

In real life you tend to see staggered subs so that there is almost always a starter or 2 on the floor at all times. In the case of Phoenix, it's unusual for Coach Monty to not have either Chris Paul or Devin Booker on the floor to run the offense. In game, they both, along with the other starters, get sat a the same time. (the run and gun Suns team I grew up watching, Mike D'Antoni always wanted one of Nash, Marion or Diaw on the floor - they wouldn't all come out together)

Is there a way to currently manage this without having to complete the matrix myself, or can the AI matrix decision making be tweaked to stagger subs/starter minutes?

[ukSUNSfan]

On a side note, the depth chart visually is fantastic this year. The style and colours make it look so much better!

[Gary Gorski]

Well the only real way to change that would be in the code, which I will look at doing, but its not always the case. If I look at this year's PHX team so far in terms of minutes the 6th highest lineup is Payne/Shamet/Nader/Johnson/McGee and 10th is that lineup with Payton instead of Payne - granted its only a few games in but its not totally uncommon.

Anyhow, I will take a look and see what I can do to make the AI try to stagger the subs so there are fewer all bench lineups.

[ukSUNSfan]

Thanks for your reply, Gary. I think the issue with looking at this season is not only the small sample size, but coming off the back of the Finals run, shortened season and Olympics, the Suns start to the season hasn't been 'typical'

By that I mean, Booker didn't feature pre-season and was on limited minutes the first few games. Rotation was higher to begin with (and probably contributed to the poor start). With Booker returning to full speed, the Suns have put together a good run. However, Ayton went down injured and the rotations have changed again (Kaminsky and McGee tend to share the minutes)

I looked at the play by play text commentary for the last game where the Suns had their 'typical' starting 5; Paul, Booker, Crowder, Bridges and Ayton - this was against Cleveland. The only limitation of the below is the NBA text comms don't appear to have included subs at end of quarters or all the timeouts. I've done my best with the below:

Q1 start - Paul, Booker, Crowder, Bridges and Ayton

05:28 - Paul and Crowder sub out (Payton and Johnson in)
03:10 - Ayton and Bridges sub out (McGee and Shamet in)

Q2 start - Paul, Shamet, Smith, Johnson and McGee

07:36 - McGee, Smith and Shamet out (Bridges, Ayton and Crowder in)
05:57 - Johnson out (Booker in)
05:26 - Paul out (Payton in)
02:58 - Payton out (Paul in)

3rd Qtr starts - Paul, Booker, Bridges, Crowder and Ayton

03:20 - Paul and Bridges out (Shamet and Payton in)

4th Qtr starts - Paul, Shamet, Johnson, Crowder and McGee

09:33 - McGee out (Smith in)
05:22 - Shamet and Smith out (Booker and Bridges in)
04:08 - Johnson out (McGee in)
00:10 - McGee out (Johnson in)

This gives an idea that it is rare for Monty Williams to not keep either Paul or Booker on the floor. The subs are staggered (and from watching other NBA games, most teams seem to make 1 or 2 subs at a time rather than platoon changes)

The depth chart in game, not just limited to the current Suns roster (on 2021 on Fictional databases the same would occur), the AI matrix almost always has 5 starters playing together and then all come out and 5 bench players come in. I'm really not asking you to micromanage the Suns roster/rotation. More can it be generally looked at for the AI to stagger subs across the depth chart instead of platooning?

[Gary Gorski]

So I did a little bit of digging into this because I really think its important to base decisions on factual numbers as much as possible in the game. Check out https://www.pbpstats.com/wowy-combos/nb ... Games=true

Last season there were roughly 3200 minutes of game time that Paul and Booker both played in the same game. Both were actually really durable last season as that was the majority of games so that's really good for this discussion and these numbers only consider games they both were playing in. I also checked their individual game logs to make sure neither of them had like a game they played 5 minutes or something and got hurt as that could skew the numbers and I did not see one for either.

46.4% of the time they shared the floor together
23.4% of the time Booker was in while Paul sat
18.0% of the time Paul was in while Booker sat

However that means that roughly 12% of the time they were both off the floor which is about 6 minutes worth of game time so while that's a small portion of the game I also would not characterize it as unusual for them to be both off the floor at the same time.

I also looked at Golden State and their two primary offense initiators last season - Steph and Draymond. Out of 2784 minutes just over 25% of the time neither was on the floor in games they both played.

So while I agree the AI can do a little better job of staggering the minutes perhaps I also do think that it's not unusual for a heavy "bench lineup" to be in and even be times where there are no starters on the floor.

[Gary Gorski]

This was addressed in b11.0.1 - will appreciate your feedback on how it looks now

[ukSUNSfan]

This was addressed in b11.0.1 - will appreciate your feedback on how it looks now

Hi Gary,

Thanks for sharing data on this! Really interesting to see the numbers behind what goes on in the NBA. The comparison in % is also fascinating between the Suns and Warriors (12% vs 25%) as that's quite a difference.

I've installed the update and had a look at the Suns depth chart and it looks much better. In 2021 game, almost always in my games it would be all 5 starters in and out at the same time on the chart. In 2022, it's the first time I've seen a depth chart stagger the starters minutes. Thanks for looking at this.

I can see my thread on team records has been closed. Just to advise having loaded up the update, the team records (not just the Suns, but the team records seem to all be from a year ago) are still not correct.

[Gary Gorski]

You would have to start a new game in order for the prior records to be updated properly

[ukSUNSfan]

You would have to start a new game in order for the prior records to be updated properly

I started a new game and the Team Records still show as they did before.

[slybelle1]

You would have to start a new game in order for the prior records to be updated properly

I started a new game and the Team Records still show as they did before.

Taking a guess here that possibly you are using the realworld mod and the place Gary changed the values is either in the default or historical database. If that is the case, I would need to upload a new version of the mod for you to see it.