OSUPiper's CH2K8 CPUvCPU Experience

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Testing the following changes from the set above:


Shooting Foul 68


Mid 48
3Pt 48
3Tend 33
Drive 78


Fatigue Rate 46
Recovery 65


*Prior to tip off, adjust each roster to a Max rotation of 9 players (see below for reasoning).


Issues:
1. I was not getting enough 3pt attempts (this only became apparent as played through a few seasons with different teams). The game engine does not provide enough opportunities for 3pFGAs outside of your main shooters, although I am getting some attempts from my stretch-4 and even a couple from a center with good range, so overall I am pleased with the change. SFs are going to heave up too many 3s, so I am watching this closely.
The interactions come from all of the tendency sliders plus the Defensive pressure setting. I can lower the Pressure (Coach) setting and lower the 3pTend. The effect will likely be (a) poor floor spacing and (b) reduced post play. I am not ready to make that trade just yet, because frankly these sliders have provided the most entertaining and engaging games I have seen in 30 seasons of this dynasty.


2. Minutes per game - I thought I had this fixed, but I was wrong. I started looking at the actual NCAA stats, and it is rare to see a team with its leading MPG under 26 mins. My sliders provide 38 min games, which means a total of 10 minutes (2x5) are missing from 200 minutes on a 5-man team IRL. So, I should expect players to average 5% less MPG in my CH2K season.
If a player averages 30 MPG in real life, they should be at 28.5 MPG in my dynasty. Fouls play a massive role in this, because CH2K8 sits players a little too long for fouls, but over all this is still a pretty sound metric. I have started messing with fatigue sliders to see if I can keep players on the floor more accurately.
The other issue is that CH2K8 does not have enough dead ball turnovers and no media timeouts. For this reason, players will sit at the scorer's table for several minutes before entering the game. Real college games have dead balls all the time.


OVERALL
It is hardly doom-and-gloom. These sliders provide much more varied attacks at the rim, a lot more jumpers, which in turn provides new outcomes due to rebounds and other interactions that did not occur as frequently. Overall, the sliders continue to improve. I still want to be careful not to favor or hamstring a certain playstyle.
The reason for the latest changes is because I watched a series of games where a superior team continued to get beat by lesser teams...which would be okay if there is a sound reason as to why (e.g., the lesser teams were able to exploit a weakness of the better team). In these games, I could not tell you why the lesser team won, outside of some foul trouble.
Consistent aberrations are exactly that -- aberrations. It is always a sign that the sliders are flawed. With these changes, so far so good, but still testing.


Anyone using the latest version posted above, I would advise making the change in this post.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

I am now editing players with a 50+ shooting tendency to have a 75+ FG rating appropriate to that tendency. I am also editing Distributor PGs to have at least decent BH and PASS ratings (75 is my initial target).


I was reading a thread in the NBA2K PC forum where I guy had created a draft class generator. This inspired me to do the test described below:


Played some games with Ranked Power 5 team hosting small school. Under my current system, I watched a small school go 0-19 from 3pt range. This has always been an issue for me. Kids do not get scholarships to go 0-19.


Put more simply, CH2K8 goes overboard with the suck level of F rated players. A kid does not get school paid for with ratings of 50 across the board. The shot distribution was good, I just think a few of those 3s would fall. When I looked into it, I noticed that two players had 3pTend over 50, but 3pFG rating was also in the 50s. All that does is hamstring the CPU, because those guys will launch 3s out of their minds and the coach just keeps throwing them out there to suck.


In this thread, I have advocated for minor edits based on player skill sets. So I found kids on a small school roster, looked at 3pTend and Drive, then raised 3pFG and Layup ratings to at least 75 for players with tendencies over 50. As a result, I watched a small school with a couple of good players (B-, C+) with slight buffs to those ratings, and the rest of the roster D-F. The B- SG dropped 30. The rest of his team scored 36 combined.


Without the edits, this kid would have scored 12-14 tops, the Power 5 Team would have been up 20 midway through the first half, and would have taken out starters, further skewing the stats. The margin stayed between 10-20 points for most of the game, and the Power 5 team was never really in any danger. Frankly, this is not just a better result, but a ton more fun and adds so much storyline to an individual game.

[Brock]

Re: OSUPiper's CH2K8 CPUvCPU Experience

I had a system for a while with getting accurate 3pt shooters: Based on this method someone suggested a while back, I lower anyone's who is C- or lower, 3pt tend down to 20 or even 0 if they're a C or PF that can't shoot; also I barely remember the exact method I came up with but it was like, 84 or higher 3pt rating got a 84 3pt shot tend, 75-83 got a 75 or something like that. I only edited the Mr. Basketballs and AA's but, I did that to make the 3pt attempts more realistic for the shooters, in hopes that if they did stay in school for 3-4 years they would probably hit around 200-300 3's in their career.

It also balances out the really good shooters from the bad ones as well.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

I can see how that system would achieve a similar result, in that bad shooters would no longer effectively break the game for CPU small schools. In doing this, did you end up with no outside shooting at small schools?


My suggested approach is to make the player ratings match tendencies, and your system would make the tendencies match the ratings. Since both systems are sound in theory, the better system is probably the one that provides the most variety with the least amount of editing.


Unless there is a more efficient way to figure that out, I suppose I will just go through some small school rosters and see what I find...

[Brock]

Re: OSUPiper's CH2K8 CPUvCPU Experience

I can see how that system would achieve a similar result, in that bad shooters would no longer effectively break the game for CPU small schools. In doing this, did you end up with no outside shooting at small schools?


My suggested approach is to make the player ratings match tendencies, and your system would make the tendencies match the ratings. Since both systems are sound in theory, the better system is probably the one that provides the most variety with the least amount of editing.


Unless there is a more efficient way to figure that out, I suppose I will just go through some small school rosters and see what I find...

I never really edit in-depth with smaller schools, only the power and major conferences. But I have implemented your older system with 3pt shooting, mainly with the smaller schools. I think it was if they're tendency was 30 or higher they needed to have a 65 3pt rating? So if they were close to 65 I just bumped them up, if not I dropped them to 20. Never kept track on how effective or ineffective the edits were though.

Edit: I just re-read your post, I can understand the match ratings to tendencies, probably more so with smaller schools, but since Choops doesn't really generate the ratings for a specific playstyle with players, I try to give them a play style with the tendencies.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

I never really edit in-depth with smaller schools, only the power and major conferences. But I have implemented your older system with 3pt shooting, mainly with the smaller schools. I think it was if they're tendency was 30 or higher they needed to have a 65 3pt rating? So if they were close to 65 I just bumped them up, if not I dropped them to 20. Never kept track on how effective or ineffective the edits were though.

Edit: I just re-read your post, I can understand the match ratings to tendencies, probably more so with smaller schools, but since Choops doesn't really generate the ratings for a specific playstyle with players, I try to give them a play style with the tendencies.

I like your logic, and I truly have not looked at whether that guy that stands out at a small school was based on FG% ratings or tendencies. I certainly believe you when you say that CH2K8 does not generate a player type, because I think that is consistent with player generation in NBA2K.


Because of how I approached the problem, I discovered a specific case at one small school that had two players above-average for that level and my edits worked just as I hoped for that small sample.


My ultimate goal would be to create a simple player generator using a spreadsheet and develop some threshold for when to use it. This would give me a consistent system with interesting results that would not add more than 1-3 edits per conference in the end.


Without a database editor for CH2K, I will not spend much time on that side of it because of the diminishing returns -- there is not enough value to this community for me to predicate a slider set on wholesale player edits (everyone appreciates min-maxing in this context, I think).

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Below is the list of sliders I am tweaking, along with current values. Not all have changed, but I have included those that I am keeping an eye on:


GAMEPLAY
Player Speed 28


Rules
Shooting Foul 68 (looking at 67)


SHOOTING
IR 38
Mid 48
3p 48
Dunk 48
Layup 35


Tendencies
Close 25
Mid 27
3P 34
Drive 76


GAMEPLAY
Screen Success 57
<table width="858" dir="LTR" border="0" cellspacing="0" cellpadding="2"><tbody><tr><td width="12%" height="20" valign="BOTTOM">Help Recovery 58
</td><td width="12%" height="20" valign="BOTTOM">
</td></tr></tbody></table>


FITNESS
Fatigue Rate 47
Recovery 65

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

A few very slight changes (in the post above). I think this is gonna be it for this set. I may tweak my in-dynasty edits to coaches/players, but I am pretty sure this is as good as I am getting the game to play.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Updated Legacy and Gametime edits on page 1 to reflect new sliders.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Dropped Drive to 75

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Version 4.0 Posted on Page One. This is likely the final slider set, although some values may change very slightly.


Version 4 is the set I am using. I believe it provides the best chance for multiple roster styles to work in CH2K8; that is, I see offensive opportunities come from all over the floor and from the players you would expect. The set has been statistically sound -- possibly skewed toward offense.


Version 3 is much more guard-oriented. If you enjoy watching more of a transition, full-court game, Version 3 may be just fine.


All slider sets are geared for use with rosters generated by CH2K8 along with my edits described on Page One. It is not necessary to edit as I do to make these sliders work; however, these sliders will produce different statistical results with user-created roster sets.

[tarek]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Hey man,

Can you explain what you mean by the following:

"c. Then, I use Random.org to generate numbers from 1-100 for every HSAA, Top 35 SF, Top 50 PF, and Top 50 C, and compare to following values:
-- For HSAAs only, PG=100, SG=83+, SF=61+, PF=21+, C=11+;
-- For Top 35 SF, Top 50 PF, and Top 50 C, SF=91+, PF=72+, C=66+;"


Is that you run random and if the number is say '87' then you edit a HSAA SG, but if the number is '74' you edit a Top 50 PF. I'm not understanding what you're doing here.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Hey man,

Can you explain what you mean by the following:

"c. Then, I use Random.org to generate numbers from 1-100 for every HSAA, Top 35 SF, Top 50 PF, and Top 50 C, and compare to following values:
-- For HSAAs only, PG=100, SG=83+, SF=61+, PF=21+, C=11+;
-- For Top 35 SF, Top 50 PF, and Top 50 C, SF=91+, PF=72+, C=66+;"


Is that you run random and if the number is say '87' then you edit a HSAA SG, but if the number is '74' you edit a Top 50 PF. I'm not understanding what you're doing here.

Yes, that is not clear at all.


The number next to the position is my threshold number. On a notepad, I list (position, player initials, and team they signed with) the HSAAs in order from #1-24, and mark the ones that are definite edits (the #1 overall, the top-rated forward, etc.). Then I write down all SFs ranked in the Top 35 Overall, and the PFs and Cs in the Overall Top 50.


Then I use random.org to give a random number to each of these players unless I already marked them for edits. I reference each number by position, so:


HSAA PGs - edit if '100' as a Superstar/One-and-Done
HSAA SGs - edit if '83' or higher, ...etc.


For busts, I am looking for low numbers (19 or lower, etc.) based on the charts.


I am generally trying to edit Up (OVR>95) about 8 players and edit Down about 4. Editing Up really gives you those Super Impact Freshmen (Anthony Davis). Some even come back for a Sophomore year even though they are clearly First Round material (Marcus Smart).


Editing Down is just a personal thing I like to add risk to that top tier and add more parity to the Power 5 teams.


Hope that makes more sense.

[tarek]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Okay that makes more sense. I have never edited so do you edit the recruits PRIOR to the season or AFTER the season once they have committed?

And to get this straight and summarize:
You edit 3 players always (#1ovr and the next 2 highest ranked position #1s).
The remaining edits are using the random numbers. So a HSAA Centre with any number between 11-100 is editable to a one-and-done? Is that correct? Is that because there are so few HSAA Centres?

Also, do you pick your busts/late bloomers the same way you select the super sophs?

And another question, does changing the heights depending on school matter? I like the idea of making the non-elite program players a little shorter, just wondering if it makes a big difference, particularly under the sim engine which you say is not the best.

ANyway, thanks for the system, i fired up the game last night and picked a team in closed legacy, so do you make all these edits straight away when you start the game? I'm really excited to have a go at this and trial the sliders.

[OSUPiper]

Re: OSUPiper's CH2K8 CPUvCPU Experience

Okay that makes more sense. I have never edited so do you edit the recruits PRIOR to the season or AFTER the season once they have committed?

After. You will end the season and advance, do all of the middle stuff, and the game will put you at Day 1 for the new season. I edit all of the Power 5 conferences and any other conference with a Top 25 team for sure, before I advance past the Preseason Show. I do this because the game generates All-Americans at that point. I edit the remaining conferences before I play against them or just as I get to them prior to conference play. (Keep in mind that I do more editing than just the HSAAs.)

And to get this straight and summarize:
You edit 3 players always (#1ovr and the next 2 highest ranked position #1s).
The remaining edits are using the random numbers. So a HSAA Centre with any number between 11-100 is editable to a one-and-done? Is that correct? Is that because there are so few HSAA Centres?

Yes, the numbers are based on several seasons of what is available on average. You will rarely see a Center.

Also, do you pick your busts/late bloomers the same way you select the super sophs?

Yes.

And another question, does changing the heights depending on school matter? I like the idea of making the non-elite program players a little shorter, just wondering if it makes a big difference, particularly under the sim engine which you say is not the best.

First, there is no fixing the sim engine; just work around it. It drives me crazy, so I re-sim a lot.

Next, Yes, height matters. I do not care who has tall players really, but small schools have too many tall players because the tall players are often really bad players to compensate. It is just how the game generates them. But height matters because the game engine takes height into account on rebounds, defense, and blocked shots. In other words, you will see the effect of height edits. It is all personal preference though.

ANyway, thanks for the system, i fired up the game last night and picked a team in closed legacy, so do you make all these edits straight away when you start the game? I'm really excited to have a go at this and trial the sliders.

One note - I do not know if or how much you can edit in closed legacy.