CPU vs CPU

[DaiYoung]

I cpu vs cpu'd the first 20 or so games of the season at default (Pro, not sure that matters) to get a feel of the issues which need addressing to make the game more closely match NBA averages.

The main problems at default were:
1) Not enough fouls, especially non-shooting fouls
2) Too many shots per game, resulting in too many points per game
3) Not enough turnovers
4) Guards (especially PG's) taking too many shots
5) Too high a percentage of PIP

Issues 2,4 and 5 are related. At default almost all teams play like the '89/'90 Loyola Marymount Lions. As is well documented elsewhere they will push it up the floor after a defensive rebound at almost every opportunity. This usually involves the PG taking it directly to the hoop almost regardless of the defense. This often results in some insane shot attempt numbers for guards, especially if they have high shooting tendencies. In the very first game I watched, Kyrie Irving (overall shooting tendency of 99) took 37 shots. As an experiment I zeroed out Irving's shot tendencies and touch tendency and he still took 9 shots! It almost seems hard coded that the cpu will attack the basket after a missed shot, regardless of tendencies, sliders or defense.

Turnovers and fouls can both be raised by slider adjustments. With all fouls at 100 (apart from shooting fouls at 90) there won't quite be enough total fouls but free throw attempts will be in line with the actual NBA. As for turnovers - the passing accuracy slider is quite sensitive. Lowering it to 35 will dramatically increase passes thrown out of bounds.

Tackling the fast break problem is much more complex. Lots of sliders potentially could help but the danger is that the half court game (which is very good at default) will be effected. My efforts so far have involved the coaching tendencies. I've halved the offensive pace numbers in each gameplan and zeroed out each teams fast break tendency.

This has helped reduce the shot totals to around about the league average - the coach profile sliders do have some effect. The fast break tendency is a very strange slider, however. From observing several games with it zeroed out what happens is that the PG will try and initiate a break as always but he will be running in slow motion. Often the defense will get back in numbers during this and the ball carrier will revert to running a play. It all looks a bit strange and I'm not very happy with it as a solution but it does dampen down the pace of play.

Getting a realistic distribution of shots between players is a tricky issue which I'm still working on. It involves editing shooting and touch tendencies (and playbooks). No clear thoughts at this stage but one player who MUST BE STOPPED is Marcus Thornton of the Kings. At default he will chuck up 30 shots a game, most of them 3's. No idea what 2k were thinking with his tendency numbers.

My sliders:

Everything default except:

Offense
Inside shot success 48
Close shot success 48
Pass accuracy 35

Defense
Driving contact shot freq 100
Inside contact shot freq 100
Steal success 35

Attributes
Offensive rebounding 40
Fatigue rate 57
Injury Freq 75 (one in-game injury in 40+ games)

Tendencies
Take 3pt shots 55
Post shots 55
Attack the basket 15
Look for post players 75
Throw alley-oops 35
Go for on ball steals 85 (trying to increase reach in fouls)
Contest shots 65

Fouls
All at 100 except Shooting at 90


Coach Profiles
Halved offensive tempo for all teams
Zeroed out fast break for all teams
Zeroed out Call Plays for all teams

[Pedsdmd]

Re: CPU vs CPU

Agreed, I am working on some sliders that are slowing correcting some of these issues

[the_future420]

Re: CPU vs CPU

Agreed as well, your thoughts on the remedies for these issues seems right in line with what I am looking at as well.

[GaryT531]

Re: CPU vs CPU

thanks for starting this thread. i use your cpu vs cpu sliders for mlb the show and you do a great job. looking forward to using these

[slimm44]

Re: CPU vs CPU

I may have found a legit fix to the fast break issue, but I ask you to help. The more testing is done, the more sure we can be of the fix.

I had tried changing difficulty levels, lowering the fast break slider to 0, too many slider variations to mention and had was seeing the same thing happen in CPU vs. CPU games - following a defensive rebound, the PG would sprint up court, do multiple And-1 Mixtape dribble moves and force up a mid-range/close shot. I can't tell you how many times I've seen Chris Paul take 26 shots in a game because of this.

I was about to just accept it and roll with it. Then, I decided something I hadn't tried yet - setting Play Calling in Coaching Profiles to 0. I thought that, maybe, this would lead to more freelance situations which would force the CPU into a different type of offense. What I've found is very surprising.

Turning Play Calling to 0 makes the CPU call plays EARLIER IN THE SHOT CLOCK and MORE OFTEN than at 100. At 100, after a defensive rebound, I've documented that the CPU waits 7-8 seconds to call a play. During that 7-8 seconds, the ball handler will usually force the action and force up a highly contested shot.

With Play Calling at 0, the majority of the time the CPU grabs a defensive rebound, a play is called within 4 seconds, they'll walk the ball upcourt and run the play. Following steals/blocks, the CPU will still look to run. For a reason unknown to me, setting Play Calling at 0 makes the CPU call more plays.

I know this sounds crazy. I wouldn't believe it if somebody else told me and I hadn't documented what I've witnessed myself. But I ask you to do this. Set the difficulty to Pro (I haven't had time to test on anything else. I assume it works on other difficulties as well but haven't had time to check.) Play with whatever slider set you want. Play with Fast Break under Coaching Profiles to whatever you want. Set play calls to 0. Start a CPU game but set at least one team to Coach Mode and turn play calling to auto, play art always and play notification always. Watch the CPU play a completely different game than what you've seen so far.

Let me know what you guys find. I can't tell you how encouraged I am by what I've seen so far.

[DaiYoung]

Re: CPU vs CPU

Thanks Peds, the_future and GaryT.

Will try that out right now slimm44. Have played briefly with Run Plays at 100 for all teams but that didn't work at all well. Maybe you are on to something with zeroing it out.

This game has so many more sliders than any other title that it makes adjustments and documenting those adjustments a Herculean (but hopefully not Sisyphean) task. The good news is that almost all the sliders seem to work.

[the_future420]

Re: CPU vs CPU

I may have found a legit fix to the fast break issue, but I ask you to help. The more testing is done, the more sure we can be of the fix.

I had tried changing difficulty levels, lowering the fast break slider to 0, too many slider variations to mention and had was seeing the same thing happen in CPU vs. CPU games - following a defensive rebound, the PG would sprint up court, do multiple And-1 Mixtape dribble moves and force up a mid-range/close shot. I can't tell you how many times I've seen Chris Paul take 26 shots in a game because of this.

I was about to just accept it and roll with it. Then, I decided something I hadn't tried yet - setting Play Calling in Coaching Profiles to 0. I thought that, maybe, this would lead to more freelance situations which would force the CPU into a different type of offense. What I've found is very surprising.

Turning Play Calling to 0 makes the CPU call plays EARLIER IN THE SHOT CLOCK and MORE OFTEN than at 100. At 100, after a defensive rebound, I've documented that the CPU waits 7-8 seconds to call a play. During that 7-8 seconds, the ball handler will usually force the action and force up a highly contested shot.

With Play Calling at 0, the majority of the time the CPU grabs a defensive rebound, a play is called within 4 seconds, they'll walk the ball upcourt and run the play. Following steals/blocks, the CPU will still look to run. For a reason unknown to me, setting Play Calling at 0 makes the CPU call more plays.

I know this sounds crazy. I wouldn't believe it if somebody else told me and I hadn't documented what I've witnessed myself. But I ask you to do this. Set the difficulty to Pro (I haven't had time to test on anything else. I assume it works on other difficulties as well but haven't had time to check.) Play with whatever slider set you want. Play with Fast Break under Coaching Profiles to whatever you want. Set play calls to 0. Start a CPU game but set at least one team to Coach Mode and turn play calling to auto, play art always and play notification always. Watch the CPU play a completely different game than what you've seen so far.

Let me know what you guys find. I can't tell you how encouraged I am by what I've seen so far.

That sounds VERY encouraging. I'm anxious to try this out later tonight.

[DaiYoung]

Re: CPU vs CPU

Okay I set play calling to zero for the next game on the schedule, Cavs-Bucks. Wasn't able to watch all of the game but here are my observations:

Total FGA in the game - 170. That's pretty good. FGA were skewed in favour of Milwaukee 94-76, mainly because they had 16 orebs. 76 FGA is probably the fewest I've seen a team take using 12 minute quarters - promising. It certainly looked like neither team was trying to break after every missed shot.

I didn't do the coach mode thing you suggested but my feeling was that the teams were freelancing much more and not calling set plays (as you'd expect I suppose). There seemed to be fewer passes, fewer picks - more simplified offense. Generally when the Bucks had the ball in the half court, Jennings would either dribble a bit and then shoot or pass to Monta Ellis who would then shoot. Very rarely did they look inside. Monta Ellis took 41 shots and Jennings 22. I don't know how much of that is down to tendencies or to a simpler offensive scheme.

I'll try it next with two team edited to provide a more realistic distribution of shots.

[DaiYoung]

Re: CPU vs CPU

Halfway through the second game with Call Plays at zero (watching more closely this time) and I'm ready to declare that you are definitely on to something, slimm44!

That annoying animation of the point guard driving through everyone and everything to get to the bucket after a defensive rebound is gone.

Updating my sliders in post one. With corrected tendencies and better playbooks I'm confident this game can be extremely close to the actual NBA now you've solved the fast break issue.

As an aside I wonder what happens with the Call Plays slider at, say, 10. Would be nice to see the CPU occasionally run and gun after a defensive rebound.

[slimm44]

Re: CPU vs CPU

Okay I set play calling to zero for the next game on the schedule, Cavs-Bucks. Wasn't able to watch all of the game but here are my observations:

Total FGA in the game - 170. That's pretty good. FGA were skewed in favour of Milwaukee 94-76, mainly because they had 16 orebs. 76 FGA is probably the fewest I've seen a team take using 12 minute quarters - promising. It certainly looked like neither team was trying to break after every missed shot.

I didn't do the coach mode thing you suggested but my feeling was that the teams were freelancing much more and not calling set plays (as you'd expect I suppose). There seemed to be fewer passes, fewer picks - more simplified offense. Generally when the Bucks had the ball in the half court, Jennings would either dribble a bit and then shoot or pass to Monta Ellis who would then shoot. Very rarely did they look inside. Monta Ellis took 41 shots and Jennings 22. I don't know how much of that is down to tendencies or to a simpler offensive scheme.

I'll try it next with two team edited to provide a more realistic distribution of shots.

It's probably due to Tendencies/Play Types/Playbooks. In the default roster, Ellis and Jennings have the following plays and tendencies. My recommended changes are in ():

Ellis

PT 1 - Isolation (PNR Ball Handler)
PT 2 - PNR Ball Handler (Mid)
PT 3 - Mid Range (Cutter)
PT 4 - 3PT (3PT)

Inside Shots 63 (52)
Close Shots 65 (35)
Mid Range Shots 98 (65)
3PT Shots 47 (47)

Use Pick - 89 (100)
Isolation - 97 (0)
Use Screen - 85 (75)
Spot Up - 85 (50)
Give and Go - 82 (50)

Jennings

PT 1 - PNR Ball Handler (PNR Ball Handler)
PT 2 - Isolation (3PT)
PT 3 - Nothing (Nothing)
PT 4 - Nothing (Nothing)

Inside Shots 69 (55)
Close Shots 50 (35)
Mid Range Shots 45 (65)
3PT Shots 82 (47)

Use Pick 95 (100)
Isolation 95 (0)
Use Screen 80 (50)
Spot Up 88 (75)
Give and Go 78 (25)

The easy way to make the gameplay a little better with just tweaking a few things are to make the changes I mentioned above. I would recommend changing more than that, and I can tell you how I would do it, but let me know. I don't want to flood you with information if you don't necessarily want it. Making those changes should show a pretty substantial difference in gameplay, though. Not perfect, but much better than out of the box.

[DaiYoung]

Re: CPU vs CPU

Thanks for that. You seem to have a very good grasp on which of the tendencies do what. Better than me anyway, I tend to go for the sledgehammer approach (for example I just zeroed out all of Mo Williams' shooting tendencies and 'touches' before this game!)

It seems to me that almost every guard in the game needs editing. Mainly lowering their shooting tendencies to stop the barrage of shots.

I notice you drastically reduced the isolation tendency for each player. How does that tendency work in the game?

[slimm44]

Re: CPU vs CPU

Thanks for that. You seem to have a very good grasp on which of the tendencies do what. Better than me anyway, I tend to go for the sledgehammer approach (for example I just zeroed out all of Mo Williams' shooting tendencies and 'touches' before this game!)

It seems to me that almost every guard in the game needs editing. Mainly lowering their shooting tendencies to stop the barrage of shots.

I notice you drastically reduced the isolation tendency for each player. How does that tendency work in the game?

It improves ball-movement and motion on offense tremendously. I had to do the same thing, last year. I actually have a 0-25-50 scale I use on Freelance Tendencies that I developed by studying synergy stats last season. Since your rosters are the default ones, it works better to use the ratings I listed earlier. I would send you my roster file but I'm on PC.

I have an editing file that is linked in a Resource Thread in the Roster Section of the forum. If you're up for editing a pretty solid amount (usually takes about 30-45 minutes per team) I can help you really shape up the roster. Do you think you'd be interested?

[DaiYoung]

Re: CPU vs CPU

Yes, certainly interested. I work from home so have a reasonable amount of spare time. I'll check out your editing tools in the resource thread. I'm on PS3 although I will probably buy the PC version at some point also.

[slimm44]

Re: CPU vs CPU

Yes, certainly interested. I work from home so have a reasonable amount of spare time. I'll check out your editing tools in the resource thread. I'm on PS3 although I will probably buy the PC version at some point also.

PM me your email address. If I get it in the next few minutes before I leave the house I'll send the excel file. I'm having trouble uploading to google docs right now. I have a really simple spreadsheet you can use.

[DaiYoung]

Re: CPU vs CPU

Okay, excellent. Have PM'd my email address.