AI Bullpen usage logic is the worst

[garry1221]

Quote: Originally Posted by garry1221

Stumbled onto this thread as I was looking for more realistic bullpen while simming games. I love team building, but i don't let my skills (lack of) dictate the games. I think i've unlocked the matrix. The look doesn't make sense, but a few months of sims are getting very pretty close, imo. I start my year with 6 Sp, 5 Rp, 1 Cp. My tested bullpen looks like: M1 - 2nd best relief M2 - 3rd best M3 - 4th best SU1 - Best/most clutch relief SU2 - 6th SP/LR/ worst reliever. Closer This set up gave me MOST of what I was looking for. I only simmed 3 months, but the only issue i saw was my SU1 had less innings than my closer. Might try simming a full season w inj off to get a real gauge of this. Using my LR/worst MR in MR 3 or 4 gave said player way too many IP in the wrong situations. With my set up, i might've lost a few by the cpu using him, even if SU1 was rested, but those were rare exceptions in my small sample size.

Small sample of testing has me slightly rethinking this post. I'm beginning to see more of a a correalation between pitching clutch and usage. Even in blowout losses, my 2 least clutch pitchers had least IP as a whole. I need to test this with said players in different places in the pen. Looking through a month, there were less than 5 headscratchers in my sim. Cpu and logic have never made total sense, but I do think i'm closer.

[garry1221]

Quote: Originally Posted by The Kid 24

Interesting... What if using 8 man pen? That seems to be the norm now.

I've never used an 8 man pen, so right now I won't begin to suggest. A bit of testing today has me more convinced that PCL has everything to do with IP. I've been ordering lowest to highest PCL, (LR2 - MR4), leaving set up empty. And closers slotted as such.

Got to thinking a few minutes ago the variables in lineups: 6,5,1; 5,6,1; 5,5,2. Using Ridin's opening day rosters, injuries removed, a couple teams had the 552 look. I want to test each set up 1st. I don't want another blown aha moment lol. I've also thought of inverting each section, so theoretically the most clutch LR is LR1/2, and the most clutch MR is at 1. I'm sure I'll test it out before too long. I've got a list of 10ish teams to test. I want to be thorough.

Thus far, the only changes i've made to my low to high clutch, are because of stamina issues. Imo, LR should have ~40 stamina minimum. If a low clutch also has low stamina, i'll drop said player to M1 and move up the 1st player meeting stamina up.

[Guys_WhoGame]

We shouldn’t have to do all this just for simple bullpen usage. I do hope they improve it next year,but at this point it is what it is. Going to finally start a franchise and just deal with it.

[garry1221]

Ok, results are starting to become consistent enough to post.

LP1 - SP6 or highest STA RP
L2
L3
M1 - best PCL
M2 - 3rd PCL
M3 - 4th PCL
M4 - 5th PCL
SU1 - 2nd best PCL
Closer

If a RP has 40+ PCL, move player to L2 and/or L3 and move up other MR. Stay in descending order of PCL. Fill LR slots 1st, then fill MR/SU in order. If running a 5-6-1 and no RP has 40 STA, just use highest STA. I've found I get most realistic results using with 1 LR, and 1 SU. I've toyed with the idea of swapping SU1 and MR1 at the ASB, but it's only a thought right now.

While going through game by game, this looked most natural given the progression of most games. Yes there were some questionable games, but it's considerable lower than other configurations. I'm not a 30 team control person, but if any of you are and want to try it out and give feedback, please do.

[bravesfan1984]

Quote: Originally Posted by garry1221

Ok, results are starting to become consistent enough to post. LP1 - SP6 or highest STA RP L2 L3 M1 - best PCL M2 - 3rd PCL M3 - 4th PCL M4 - 5th PCL SU1 - 2nd best PCL Closer If a RP has 40+ PCL, move player to L2 and/or L3 and move up other MR. Stay in descending order of PCL. Fill LR slots 1st, then fill MR/SU in order. If running a 5-6-1 and no RP has 40 STA, just use highest STA. I've found I get most realistic results using with 1 LR, and 1 SU. I've toyed with the idea of swapping SU1 and MR1 at the ASB, but it's only a thought right now. While going through game by game, this looked most natural given the progression of most games. Yes there were some questionable games, but it's considerable lower than other configurations. I'm not a 30 team control person, but if any of you are and want to try it out and give feedback, please do.

Very intrigued. Can't wait to see more results. Looks like you may be on the right path. Kind of had a feeling clutch played some role in this.

[The Kid 24]

Quote: Originally Posted by garry1221

Ok, results are starting to become consistent enough to post. LP1 - SP6 or highest STA RP L2 L3 M1 - best PCL M2 - 3rd PCL M3 - 4th PCL M4 - 5th PCL SU1 - 2nd best PCL Closer If a RP has 40+ PCL, move player to L2 and/or L3 and move up other MR. Stay in descending order of PCL. Fill LR slots 1st, then fill MR/SU in order. If running a 5-6-1 and no RP has 40 STA, just use highest STA. I've found I get most realistic results using with 1 LR, and 1 SU. I've toyed with the idea of swapping SU1 and MR1 at the ASB, but it's only a thought right now. While going through game by game, this looked most natural given the progression of most games. Yes there were some questionable games, but it's considerable lower than other configurations. I'm not a 30 team control person, but if any of you are and want to try it out and give feedback, please do.

Any updates? Would you recommend setting up bullpens like the above?

And curious how you think 8 man pens should be set up, use SU2 or LR2 spot for the 8th guy?

[Great Whale]

The AI logic cannot be fixed.
What I do is simulate (fast-forward) each game,
with these rules:

*Stop Game when one team scores 8-runs; When 8th Inning Starts; and I do all pitch changes from 9th Inning onward*.

This will remedy many issues. True, I play this game like it is OOTP, and only play when I bash away on Beginner Level, in Retro Mode, LOL.

Best idea is to set up 2-teams and go into play now.
Use a great team vs a bad team.
You can know fast-forward and simulate many games, to test Cpu AI Logic. I have done this with 100s' of games.

Seems the LRP #1 is the guy uses in blow-outs, and 2-IP in close games when your team us behind (6th and 7th Innings).

SU #2 is the spot used least often by Cpu AI.
Problem with putting best MRP in MRP #1 slot, is he will eat up some useless IP, and be skipped for 8th Inning, Holds Situations.

I don't think Clutch, or Stamina means a whole lot, really.
I think the name of the BP slot (LRP - MRP - SU - CL) dictate what pitcher is used most

Good thread.

[Great Whale]

If you want to manage each game,
via fast-forward, try this set-up:

*LRP #1 - 6th Starter, w/ Stamina @ 60.
*LRP #2 - Worst Overall RP, w/ Stamina @ least 20.

*MRP #1 - 2nd Best RP.
*MRP #2 - 3rd Best RP.

*SU #1 - Best RP.
*SU #2 - Lefty-Specialist.

*CP - Closer

The Stamina Scales I use are:

Starters: 80 (Default) - 90 - 99.

Relievers: 10 (Lefty Specialist)
20 (MRP)
30 (2-IP, MRP)
40 (Closer)
60 (6th Starter, LRP #1)

You can also increase Reliever Stamina up to maximum of 10, when LRP gets called into game. I suggest that if you really want realism, you must fast-forward each game, and manage blow-outs.

Also, for more accurate Closer Saves totals, remove CP tag for initial 25-Calendar Days of Season. Put real closer as MRP #1 slot, and Label him as RP. Label real MRP #1 as CP, and insert him in CL Slot.
Put MRP #2 in SU #1 slot.

At end of 25-Calendar Days, revert to normal, as I have listed above. Also, you can swap the spots of LRP # 1 and 2, as well MRP # 1 and 2. Do this @ start of each new month. This will scatter IP better for LRP, AND Holds/IP better for best 2-MRP.

Is it worth it ? Hey, for us baseball fans, it surely is worth it.