AI Bullpen Logic

[ripwalk]

Re: AI Bullpen Logic

After a fair number of simmed seasons … I think I’m going to go with these as my simming sliders:

Human/CPU Starter Stamina 7
Human/CPU Reliever Stamina 2
CPU Manager Hook 7

Still need to do more testing … but at least for me it gave an appropriate amount of SPs over 200
IP, and there were still some RPs with insane amounts of innings, but it was more like 2 or 3 across the league versus every team having an 150 IP reliever.

But the majority were in a more realistic range that I can live with.

I’d be curious to see what results other people are getting and who is going with what.

[KnightTemplar]

Re: AI Bullpen Logic

Looking at “current” rosters in the vault I see a lot of SP who will be pitching out of the pen this year are still SP with SP stamina ratings. Do you guys edit them to RP with the appropriate low stamina, or roll with the roster as is?

[MiracleMet718]

Re: AI Bullpen Logic

Looking at “current” rosters in the vault I see a lot of SP who will be pitching out of the pen this year are still SP with SP stamina ratings. Do you guys edit them to RP with the appropriate low stamina, or roll with the roster as is?

I’ve left it as is and it seems like they get some starts and pitch a bunch of innings out of the pen, but they do get used as starters for a few starts and/or sent down until a spot opens up (usually around 130-140 innings total at most). I think based on my findings, that kind of checks out to what real life shows. The issue for me continues to be that no matter what you do for RPs and stamina, they don’t send down/call up enough pitchers to make up the difference in innings.

I’m going to also mess around with durability attribute to see if maybe guys get some minor injuries and see if it improves the send down/call up issue. However I don’t think that will solve it either. I think SDS expanding the rosters to 120 would help since they could put in more logic to change roster logic and use more RPs. Until then we’ll have to do with a few RPs being in the 90+ innings and saves being too high, but most having appropriate innings.

[KnightTemplar]

Re: AI Bullpen Logic

I’ve left it as is and it seems like they get some starts and pitch a bunch of innings out of the pen, but they do get used as starters for a few starts and/or sent down until a spot opens up (usually around 130-140 innings total at most). I think based on my findings, that kind of checks out to what real life shows. The issue for me continues to be that no matter what you do for RPs and stamina, they don’t send down/call up enough pitchers to make up the difference in innings.

I’m going to also mess around with durability attribute to see if maybe guys get some minor injuries and see if it improves the send down/call up issue. However I don’t think that will solve it either. I think SDS expanding the rosters to 120 would help since they could put in more logic to change roster logic and use more RPs. Until then we’ll have to do with a few RPs being in the 90+ innings and saves being too high, but most having appropriate innings.

I’ve read where guys complain they bring in the same pitchers in relief most of the time. I’m wondering if the AI looks to stamina to determine who gets used first/second.

I’ve left only one man in SU. Last year, with 2, it seemed one of them might never get used. So far, only about 16 games I’m seeing every pitcher getting innings.
At least they’re being used.

[bfindeisen]

Re: AI Bullpen Logic

Hahaha, I cannot believe that these issues are still so prevalent. At least 10 years now. The amount of contorting and manipulating and flat out work the ppl in these threads do year after year in an attempt to fix these issues is astounding, yet they continue to persist. I truly don't know where ppl get the motivation to keep coming back and working the same problem. I gave up years ago and just check in ever year now to see if any improvements have been made at all, and they never are.

[KnightTemplar]

Re: AI Bullpen Logic

Hahaha, I cannot believe that these issues are still so prevalent. At least 10 years now. The amount of contorting and manipulating and flat out work the ppl in these threads do year after year in an attempt to fix these issues is astounding, yet they continue to persist. I truly don't know where ppl get the motivation to keep coming back and working the same problem. I gave up years ago and just check in ever year now to see if any improvements have been made at all, and they never are.

Well, we have 2 choices:

Look for a workaround and make the best of it because you love the game.

Quit playing, as it appears you have, because it’s a game breaker.

[MiracleMet718]

Re: AI Bullpen Logic

I’ve read where guys complain they bring in the same pitchers in relief most of the time. I’m wondering if the AI looks to stamina to determine who gets used first/second.

I’ve left only one man in SU. Last year, with 2, it seemed one of them might never get used. So far, only about 16 games I’m seeing every pitcher getting innings.
At least they’re being used.

I’ve only run into that once where the Mets had Tylor Megill sitting as an SP with 0 innings, but he was a late season call up used to replace an injured starter in the last week. Otherwise all other pitchers are being used. I’m only using one team control and letting the CPU control everything as part of these tests.

The usage is definitely an issue though. When playing games last year, the CPU would always throw in the same relievers when their team was losing and an SP got run early. Those guys would stay in until they got tired or also got knocked around. I think it’s fine if they would just move more pitchers up and down though.

Like I said the game is using half the amount of pitchers actually being used by MLB rosters, so it’s forcing the guys being used to pitch more innings than they should.

[soxfanbs91]

Re: AI Bullpen Logic

Looking at “current” rosters in the vault I see a lot of SP who will be pitching out of the pen this year are still SP with SP stamina ratings. Do you guys edit them to RP with the appropriate low stamina, or roll with the roster as is?

I'm wondering if it makes sense to lower the guys who are SP in long-relief spot's stamina to like 40, or even less. I feel like any long-reliever who makes a spot start doesn't last too long into games anyway, going like 4-5 innings or even less than that.

[djflock]

Re: AI Bullpen Logic

Hahaha, I cannot believe that these issues are still so prevalent. At least 10 years now. The amount of contorting and manipulating and flat out work the ppl in these threads do year after year in an attempt to fix these issues is astounding, yet they continue to persist. I truly don't know where ppl get the motivation to keep coming back and working the same problem. I gave up years ago and just check in ever year now to see if any improvements have been made at all, and they never are.

It is what it is at this point, got to try and make the best out of it. Just hate when someone on here tells me "its not real life, its your universe"

[MiracleMet718]

Re: AI Bullpen Logic

I'm wondering if it makes sense to lower the guys who are SP in long-relief spot's stamina to like 40, or even less. I feel like any long-reliever who makes a spot start doesn't last too long into games anyway, going like 4-5 innings or even less than that.

I bring this up again because even if you do this, how do you solve them not using enough pitchers to begin with? There’s usually 23-30 guys that pitch out of a bullpen per year whereas in the show, it’s only 10-13.

If we use the Mets as an example, they had 558 innings from pitchers in relief across about 27 different pitchers (including guys who started some but also pitched as RPs). So that’s roughly 21 innings per reliever vs 50 innings per reliever in The Show.

[KnightTemplar]

Re: AI Bullpen Logic

I'm wondering if it makes sense to lower the guys who are SP in long-relief spot's stamina to like 40, or even less. I feel like any long-reliever who makes a spot start doesn't last too long into games anyway, going like 4-5 innings or even less than that.

Well, all I can say is I go thru all the rosters, scroll down, and any pitcher who has high stamina I reduce to around 29-30. Something I’ve just done. And one set up man. So far it’s working for each team, meaning different pitchers do look like the get their work in. I found if you leave an SP with high stamina in an RP slot he’s gonna eat up a ton of innings.

Nothing is going to be perfect, but it appears better so far.

[MiracleMet718]

Re: AI Bullpen Logic

This is probably the last I’ll post about the transactions part since it’s kind of useless to keep mentioning it as we can’t really fix it. However I changed the pitcher durability to around 30 for all Mets pitchers hoping I’d see more injuries/moves up and down and I did see that a bit (average of 20 pitchers pitching in a season vs the 15-17 I was seeing before).

However I checked the transactions and compared to the Mets pitchers moves last year. In my franchise, I saw an average of 10 moves all season. In real life, the Mets made 17 pitching moves in April alone (just options/recalls, not including waivers). So I think this just proves to me that player movement/roster size is the actual culprit here.

Still working with tweaking stamina to see if I can get the innings numbers down a bit, but it’s going to be a case of “best we can do” rather than actually making real progress. That is of course until they expand the rosters from the 93 they give us and program in more options/waivers/recalls into the game.

[BIGBLUEFAN2023]

Re: AI Bullpen Logic

Well, all I can say is I go thru all the rosters, scroll down, and any pitcher who has high stamina I reduce to around 29-30. Something I’ve just done. And one set up man. So far it’s working for each team, meaning different pitchers do look like the get their work in. I found if you leave an SP with high stamina in an RP slot he’s gonna eat up a ton of innings.

Nothing is going to be perfect, but it appears better so far.

I've done this as well and it's not perfect but much better IMO. The LR still goes 3 innings or so when brought in early (2nd or 3rd inning) but they are not coming in as much in close & late situations.

[Christian9]

Re: AI Bullpen Logic

Sorry fellas, been a busy week so far. And having a 4 y.o. at home doesn't always make life predictable.

I've run a few additional scenarios and some basic notes below. Wish I had more time to summarize or really dig in, but haven't lately. Some below are rhetorical, and some nonsense, but wanted to note them as I went along on last few scenarios.

Please let me know what else I can run, and I can likely get right around to them. Sorry if you asked last couple of pages, haven't caught up on chat.

https://docs.google.com/spreadsheets...it?usp=sharing

Trying on Google Docs (hopefully link works) as well.

To limit CGs to reasonable amount, CPU Manager Hook (CPUMH) needs to be at least 8 (with no Stamina slider changes) or Stamina at 0 seems to fix this, but with testing CPUMH at 5 there were too many GP (appearances) for RPs. CGs quickly gets out of hand if not purposefully mitigated.

If no CP, SU1/SU2 with higher PCLT has more SVOs.
MRP1 - 2nd most appearances or most HLD & SVOs? SU1 or SU2?
LRP1-3 prioritizes Stamina for IP? All average > 1.1 IPG?
Higher stamina in MRP1-4 does not seem to matter.
MRP1-MRP4 prioritizes PCLT.

I'm really looking for insight from you guys. A fresh set of eyes!

[MiracleMet718]

Re: AI Bullpen Logic

Sorry fellas, been a busy week so far. And having a 4 y.o. at home doesn't always make life predictable.

I've run a few additional scenarios and some basic notes below. Wish I had more time to summarize or really dig in, but haven't lately. Some below are rhetorical, and some nonsense, but wanted to note them as I went along on last few scenarios.

Please let me know what else I can run, and I can likely get right around to them. Sorry if you asked last couple of pages, haven't caught up on chat.

https://docs.google.com/spreadsheets...it?usp=sharing

Trying on Google Docs (hopefully link works) as well.

To limit CGs to reasonable amount, CPU Manager Hook (CPUMH) needs to be at least 8 (with no Stamina slider changes) or Stamina at 0 seems to fix this, but with testing CPUMH at 5 there were too many GP (appearances) for RPs. CGs quickly gets out of hand if not purposefully mitigated.

If no CP, SU1/SU2 with higher PCLT has more SVOs.
MRP1 - 2nd most appearances or most HLD & SVOs? SU1 or SU2?
LRP1-3 prioritizes Stamina for IP? All average > 1.1 IPG?
Higher stamina in MRP1-4 does not seem to matter.
MRP1-MRP4 prioritizes PCLT.

I'm really looking for insight from you guys. A fresh set of eyes!

Very interested in your findings. I’ve been posting some things as I go, but I need to do more of my own analysis. I do know for SPs, I had a lot of success for lowering CGs and getting a handful of 200 IP pitchers by using SP Stamina = 1 or 2, RP stamina = 0, and Manager Hook = 8. Been fooling around with that more along with the injury slider to see if that helps get more roster moves so more minor league pitchers come up and eat up some RP innings, but not much success.