OSUPiper's CH2K8 CPUvCPU Experience

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

For those of you watching at home, things are improving slightly. These sliders are surprisingly sensitive.


Presently testing the following changes from the set on Page 1:


Midrange (FG%) 45


Close tend 15
Mid tend 62
3pt tend 18
Drive 65


Screen Success 48


Help D 31
Help Rec 47

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

It is nuts.


My guess is there are 'sweet spots' created at certain ratios of sliders that interact (tendencies and Help D/recovery), and even just on one slider (ballhandling). As an experiment, you can adjust the 3pt tendency slider or the drive slider, then watch a guy with relatively equal tendencies to shoot a 3 or drive (this is what I did). You really see the sensitivity as one slider moving one spot will completely tip the balance on how that player plays.


The FG% sliders are similar. Move the Layup slider down 1, and watch how many more blown layups there are.


The butterfly effect over the course of a game is pronounced, especially if "6th Man" momentum kicks in.


After these changes, I watched Duke lose to an 0-7 Princeton team and a mediocre Vanderbilt team. In the legacy season, Duke is a jump-shooting team, a few athletic guys who can make plays at the rim, but woeful in the post. These changes were intended to create more contested 3s, and it seems that is happening. So I bumped 3pFG% up one point. If this generates 1-2 more makes per game, out of 15 attempts, that's a 12-15% increase in 3pFG%. Plus, a well-timed 3pointer may push the momentum to the red, leading to a steal and a run-out basket, a defensive stop and another 3, and all of a sudden, that 1 point increase in your slider generated an 8-point run in a game.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Close Tendency back up to 16


Screen Success up to 50


Happy with the results at this point and very close to releasing these. I like where the numbers are and where shots are coming from, although I would like a little more post presence and about 2-3 FGAs less per game (per team) overall.


I am hesitant to lower the Mid Tendency because I am afraid the midrange game will disappear, but I do think that is where those extra shots are coming from. Raising Drive by 1-2 might be interesting to look at, especially with the bump to Screen Success. My thought is that would lower FGAs slightly, increase attacks on the basket in the two-man game, and create a few open looks from 3 (all of a sudden, open 3s are hard to come by with the recent testing changes).


If anyone is trying the numbers I have been posting this week for testing, feedback is welcome. Feedback is welcome on any part of the set or on your own modifications, for that matter.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

i have been testing the updates, but using the setup mentioned in previous post--dartmouth, quick game. i have noticed better shot distribution, with improved determination in the low post-i.e. post players receive pass, commit to the shot rather than look to kick out. have also noticed more fga per game...with original set, both teams had attempts in the 50's, with update, at least one of the teams is in the 60's. i would think that this would put a team in the upper tier in fga in the legacy team stats. i understand the quandary-lowering shot tendency throws off the flow of the game, reducing minutes lowers stats in other areas.

i tested upping close tendency to 75, with other tendencies relatively lowered in trying to emphasize low post, yet keep overall shot attempts down. however, this created kind of like a fast paced princeton effect, with both teams back-dooring like crazy, and making court long passes on the break. the pace alone counteracted the lowering of other tendencies, resulting in high fga and unrealistically high points in the paint.

i will say that the current setup accomplishes things that i have never seen in other sliders--consistently substantive offensive rebounding numbers, and a more random, organic use of the shot clock. i also agree with your theory of the "sweet spot" in sliders at which the game engine becomes less patterned and more contingent upon circumstance.

I also noticed FGAs rising, but I tested against a couple of fast-paced offenses and was not concerned with that issue.


As an added wrinkle, a Coach's Tempo rating seems to be overshadowed by the players' tendency ratings. Because most low-rated players have correspondingly low shot tendencies, it is ineffective to take a Westhead-LMU-run-and-gun speed and apply it to a small school. Likewise, I cannot get a major program with good talent to slow the game down. I have no fix for this.


Lowering Off Awareness reduces shot attempts and rocket passes to the baseline; however, lowering Off Awareness will also reduce the number of 'improv' dribble-drives, cross-over animations, etc. I have not been willing to make that trade. Thus, I am left with the more probable step of reducing some or all tendencies to artificially slow the game down, then slightly raising the Shooting Foul slider to get fouls and FTAs back in line.


When testing CLOSE=75, where did you have other tendencies?

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

i have played a lot with lowered offensive awareness, shot selection suffers, offensive options become simplified and predictable. i would think that if slightly elevated fga is the only unintended by-product, it is an acceptable trade-off.

100% agreed. Off Awareness must stay high. Raising Def Awareness is acceptable until drives become ineffective.


This is the struggle IMO at this point. For example, I just dropped all tendencies by 2 to watch the effect on FGAs. The result was as follows:
Total FGAs = 106; Total FTAs = 37.
These numbers are almost exactly where they should be. Additionally, both offenses ran more sets / more designed plays. We already know that more shots out of set plays results in better shot distribution, so all of these results are good.


The problem is (1) lack of dribble drive (even with these slight adjustments and (2) a slight but distinct advantage to the team with more second chance opportunities / fast break opportunities.


As to the second problem, this is a fair result, as the ability to score in transition and second chances is a major advantage in basketball.


The first problem is trickier; I have been watching games with my DRIVE slider as a focus for several seasons now. It could be that seeing fewer drives reveals my own bias as an observer (that is, I see fewer drives as a problem even if it is not). My goal is not for every player to rocket into the lane, but rather for those athletic players who attack and play above the rim to affect the game appropriately.


Fortunately, I have a perfect 6'5 swing man to watch and see how this plays out. Will watch one more game tonight on this setting and report.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Consider these numbers, pulled from my present legacy. Most of these games were played with the sliders on Page 1, with less than 25% played with the recent changes to Help D, etc.


Conference FGAs
Max = 58.58
My Games = 58.44
Min = 47.25
Median = 53..22
Avg = 53.21


FTAs
Max = 35.07
My Games = 22.06
Min = 22.06
Median = 25.79
Avg = 26.90


FGAs + 1/2 FTAs
Max = 73.75
My Games = 69.47
Min = 59.29
Median = 67.95
Avg = 66.66


"FGAs+(FTAs/2)" is a dirty shorthand for offensive possessions with an attempt at the basket. Note that I observe one of the highest FGAs/game, but the lowest FTAs/game. I am about 3 possessions/game over the average for a power conference.


Also consider that it is impossible to create enough turnovers to match the population, statistically, so all of these other teams actually have more possessions per game than in watched games. In fact, I see about 3 fewer TOs per game than the conference average.


I agree with you, jww, in wanting FGAs per game to be lower; however, this numbers tell me that I am almost spot on in possessions/game, but far behind in FTAs. Part of this is a lack of inside scoring, which means fewer contested shots or shots in traffic, which means fewer trips to the FT line. Secondly, I have a jump shooting team this season. Given all of this, I would expect slightly above average number of possessions (and FGAs) and slightly below average FTAs. I would probably expect to see a few more turnovers (1-2 per game above where I am).


So, I am going to (1) add back in those 2 pts to each tendency; (2) lower ballhandling slightly (for more TOs); and (3) increase Shooting Foul. (NOTE - I have been playing with Reaching Foul and Shooting Foul, so those numbers may differ from the Page 1 post.)

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Dropping Ball Handling to 28 did not appear to increase TOs but did negatively affect drives. I may test raising Def Awareness by at least 5 points to see the effect on drives. I am hoping that Help D settings will keep the lane open. If so, that may be a direction I can go. It should even out 3pFGAs.


When DRIVE is lowered, jump shots are more evenly contested. This is great for producing stats, but not nearly as interesting as the effect of a good dribble drive. Since a good penetrator is such a commodity in this game, I want to maintain the great impact it has on the gameplay. Basically, I think the effective Drive is the key to 'randomness' in this game.

[Brock]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Been late to ask this: What made you decide to drop team unity by 10?

edit: Also, I've seen some of the NICEST blocks with your sliders.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Been late to ask this: What made you decide to drop team unity by 10?

edit: Also, I've seen some of the NICEST blocks with your sliders.

Man, I know. The action on the ball is awesome. VC made such a great game, it is funny that we are still tapping into new stuff.


As for Unity, I was afraid that my sliders could never support a Fab 5 or Kentucky team of 1&dones. Frankly, I did not see impact Freshmen like I wanted to. In retrospect, I can't tell if it made a difference.


Right now, offense is too aggressive and effective. Of course, games are more fun to watch than most real life contests, so I am having problems generating enough courage to start throttling the tempo down.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

I think the answer is Defensive Pressure in the Coach settings. Not sure why I did not think of this.

Consider the issues: not only are FGAs are trending high, yet still inconsistent, but 3pFGAs and post play are even more glaring. This is because of the different degrees of pressure from different teams.

This has always bothered me. IRL, if you are getting torched by someone, you adjust. If you are in foul trouble, you might sag on D. In CH2K8, coaches cannot react like that because of the preset Pressure rating for that game.

Really what we want to accomplish is good spacing and contested shots when the offensive player wants to play Hero Ball. In contrast, we want to reward ball movement and superior one-on-one skill.

I have DEFENSIVE PRESSURE set to 90. It helps spread the floor, which plays a huge factor in the great results we have been seeing.


Although 90 is high, and like you would expect, defenders are right in the jock of their guy, the flow and shot selection looks great. I may tinker with dropping Pressure to 85 or 80, but so far so good.


This strategy makes sense. It provides uniformity from game to game, which will in turn promote statistical results based more closely on team composition rather than favoring a position. Likewise, tendency sliders will be more effective and respond in a more predictable way. FG% sliders will also be easier to set, meaning open jumpers/layups can have high FG% without breaking FG% overall.


I highly recommend implementing this change.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Quick note for those following this thread: raising Defensive Awareness had minimal impact on shot selection or offensive flow.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

so are you using the latest updates, i.e. help d/recovery at 31/45, screen success at 50, close tendency at 16 and team defensive pressure at 90 as the only changes to the main post slider set?

Yes, I believe all of those are correct. I have been messing with tendencies and D awareness today, but have not seen improvement.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

im testing your latest update, but using the x/2+50 formula for defensive pressure...keeps some variation but maintains a higher threshold of aggression for every team. looks pretty good so far.

Thanks, I am going to try that as well.


I am testing Defensive Pressure at 60-90 with good success.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Just finished a season where my team was loaded at positions 1-3. Had a good PF, but nothing else inside. Was playing 6'5" at PF with my 6'7" post playing out of position at C. We were a jump-shooting team with the exception of a backup PG and a lottery pick at SG (playing SF) who were 1-on-1 threats. As you would expect, we were up and down, winning 7 games against the RPI Top 50, but losing 3 games outside the Top 100. Advanced to the Sweet 16 and fell 4 points short of the Elite 8.


Last season I had recruited well, landing the #2 SG in the nation and a 4-star SF to replace my backcourt. Unfortunately, I just did not have any minutes for them. After the NCAAs, I find out that BOTH of these guys are transferring out. Frankly, I don't know if I can recover from that in the ACC. Honestly, I almost bolted for the Pac-10.


In other news, sliders are holding up... I will test these last changes through several games this week and likely release them. (NOTE: In browsing some old 2k8 threads, I noted a theory that the "Speed" attribute setting relates to a speed burst that a User can execute with the gamepad. Think of it as modifying the quick first step on a drive. I am going to try this out at 100 and see what effect it has on drives to the basket. If it works, I think this is a major piece in balancing shot distribution.)

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

I am currently testing these particular slider changes (below), based on putting Speed=100, to see the effect on drives to the basket. These appear to have reduced FGAs substantially. I like how they look, but not sure I like them better than what we had.


It does seem like the boost to Speed helps the ballhandler get an angle on the defender. I need more games to make any real judgment. If anyone is interested, here are the changes:


Game Speed 42
Player Speed 28


Close 15
Mid 57
3pt 19
Drive 68


Off Aw 70
Def Aw 75
BallH 27
Speed 100


Help D 30
Help Rec 46