ACE & Pace!

[antdoggydogg]

Suggested Reading: (Questions and Answers from the Czar AI thread)

Spoiler

Quote: Originally Posted by ronyell

How will the engine determine if a team is fast or slow paced years into a MyLeague or if the team composition changes in regard to tempo? How will ACE take into account those same situations. How will these AI enhancements play out in tandem with the new coaching philosophies mentioned in the other AI blog?

Quote: Originally Posted by Da_Czar

For this first year there isn't a tie-in between system proficiency and ACE. Right now they are two separate systems. But your line of questions would definitely seem like a natural progression. What I'd like to know from you is what criteria would you like to see considered for a more dynamic tempo adjusting AI in terms of coach philosophy etc ? Maybe start another thread on that so I can find it if needed.

Quote: Originally Posted by Da_Czar

For pace my stats guy grabbed the end of season pace of play for all teams and we adjusted the coaches ACE or adaptive coaching engine accordingly. Because it is dynamic we don't have absolute control over it. so if a slow tempo team get a ton of stops obviously they will be in transition more and they could have inflated scores. The hardest teams to regulate is that sloth of teams right in the middle it is hard to differentiate between the teams that score 98 and the teams that score 99 right ? So that is why in the blog I mentioned you will see a difference between the top and bottom teams in terms of pace. So it's not an exact science but we put some time into it and I verified by watching a ton of AI vs AI this year to see how the scores shook out. There are still more designs I have for how to improve this in the future to give me more granular control over tempo. If you have suggestions throw em out on what you would like to see or control.

Quote:

Mostly just the slower teams. This is another one of those it's a game not real life scenarios. To get the higher scoring teams to score above 120 we couldn't allow them to walk it up becuase on average they would walk it up too many possessions. I don't have that type of granular control yet to allow them or limit them to x amount. The only way I could get their AI vs AI scores up with to allow them to run it up. So on the top scoring teams you will barely see them walking and the slowest teams will walk alot if the opposition is scoring on misses everyone runs.

Quote:

I like how you are thinking, at this point no we do not have that but I added your suggestion as something to think about. What type of variance would you expect if Deron was on an uptempo team ? Is the coach more important than an individual players tendencies in regards to pace ? Is there a big enough difference between a low walk up tendency player on a fast team and a fast walk up tendency player on a fast team to justify the work of putting that in ?

Quote: Originally Posted by trekfan

I would say that the individual players tendency would take precedence -- players play, coaches coach, and players got to buy in. On the second question, ideally, a slower player like Deron Williams wouldn't be playing in a fast paced system that required him to bring the ball up -- Williams' system proficiency in such a scheme would be low, so the coach would deploy him differently/vary the system if he were playing. I don't think the system proficiency feature is even close to grasping that kind of detail yet, but in the future you wouldn't see someone like Williams running a run and gun system as a first option -- the coach wouldn't start him in that and the AI GM wouldn't sign Williams for that.

Many thanks to Czar for engaging with us in this dialogue. Along with Ronyell and TrekFan I was also very curious about the pace that teams play at. I'm a freak when it comes to getting gameplay and simulated stats looking true to life. Here are my ideas on possible solutions.

From what I understand, each Coach has their own unique ACE preferences, and the part that controls pace is based off of last seasons Pace stats. That's good.

I think we can all agree that for the most part, every coach in real life has their own identity, and when they switch teams they usually carry over their key philosophies, some more successfully then others. A lot of there success/failure has to do with System Proficiency (now in the game!), which is very much a real thing. Some Coaches are lucky to have rosters that match their preferred System. Some end up with a Carlisle and Rondo situation, where their starting point guard doesn't have the shooting skills that the coach's system requires

Or D'Antoni with the Lakers, where they had an older team that couldn't run as much with two Centers in the lineup that didn't provide enough spacing for his system to function properly. The conflicts between roster lineups, and system proficiency will develop on their own in these situations because you won't be getting the most out of said system.

So with said, there should definitely be a tie-in with ACE and system proficiency in future games. I think Coaches should have their preferred ACE tempo, which can be based off their coaching career's statistical track record (attached an image of my own research below as an example). Some like to play fast, and some like like to slow it down, while some don't really have a preference on that aspect of the game. These should work in conjunction with their preferred system.

Same goes for players, some like to run up the court while others are more slow paced. Although if you really think about it there aren't THAT many current examples of point guards who slow the game down and walk it up a lot. Off the top of my head I can think of Mike Conley. I also recall nominal "point guards"/ball handlers like Kobe, Wade and LeBron walking it up after made baskets.

Now that both the Coach and Player have their own preference/tendency, it brings the game even more life-like scenarios and challenges. It builds off of the system proficiency features that were added this year. It's something I've wanted for a long time, and now you can add this extra wrinkle. The clash between player and coach philosophies should exist and could be controlled by some sort of "Coach-ability' tendency or attribute. Being coachable is more of a skill, while the actual pace preference is more of a tendency but I worry that introducing pace tendencies to each player will mess with simulated MyLeague Stats heavily. So maybe their speed and athleticism controls the pace they play at while their coach-ability attribute adjusts their ability to adapt to the coach's system. This might not even need a new attribute, it could be lumped in with "Intangibles".

Figuring out how to make the tempo dynamic to each individual game is tricky. Cause ideally all these coaches usually stick to their usual game plan and identity, but against certain opponents in the playoffs they rework things and make adjustments. I think dynamic tempo changes should be saved for playoff series adjustments. Similar to how the Cavs purposely slowed down the game against the Warriors during their first two Finals meetings.

It's totally normal for a team to push the pace after getting stops or turnovers, if the fast break opportunity is there most teams take it. So that never bothered me because it was only natural.

In 2K17, I was never able to differentiate what ACE settings the different teams were playing on. Because I rarely ever saw them walk it up the court. I felt like they all adjusted to play at an average pace as soon as the games started regardless of the opponent. I don't mind when ACE starts strategizing for certain players defensively and adjusting their defensive focus. But they were too quick to adjust their ACE settings in regards to the basics like offensive rebounding/defensive rebounding philosophies, and Pace.

I'm impressed by how things look in 2K18 and this issue looks corrected. Coaches who spend entire training camps and seasons trying to get their team to play at a certain pace or leak out in transition/Crash the defensive glass don't just abandon those core philosophies against every regular season opponent. This should be more rare, like against very specific teams that are a bad matchup for them.

I don't even think its entirely a points per game issue. As long as the correct number of shot attempts are there I'm cool with whatever the final score is. Because maybe a team that normally scores a lot of points got their usual shot attempts but faced a tough defense that night. The highest scoring average last year was 116 by the Warriors, and the league average was 106.

Basically, I'd like a coach's ACE adjustments to be in-line with their system proficiency/POE preferences. If we could have control over each coach's ACE PREFERENCE like we do with POE's, I think that would give us more control over how these teams play, and allow us all to find our own happy medium. I don't know if this is something thats possible because ACE sounds like a very smart and complex system.

I wish there was a way to find out how many shot attempts are averaged in AI vs AI games by the teams that are set to "walk it up". As a user I would be able to test this out and figure out what teams I want to assign that ACE setting to.

Sorry for the long post, or if it seemed like I was rambling. I had a lot of thoughts on this stuff and theres never a quick easy solution.