MLB The Show 16 Bullpen Tuning Details in Patch 1.04

[EWRMETS]

Re: Bullpen tuning in patch 1.04

I think it's great a closer would have to get some work but is there a penalty if a reliever goes a long time without pitching?

[Brian SCEA]

Re: Bullpen tuning in patch 1.04

CG's have been very, very close to looking like real life numbers with this setting.

I'll take a look at CG stats in general next time I have a chance. The stat depends heavily on subjective "borderline" criteria.

The main focus of my changes were on IP/G and the fact that every starter would go an extra half inning per game. That adds up and heavily affects SV and G for relievers and also ends up potentially hiding other logic issues. The original problem was parts of this simming logic is shared for in-game manager decisions, so I had to fix things regardless of if I actually wanted to, in order to address in-game issues.

In other words, in-game logic was pulling starters too soon for one set of reasons, sim logic was letting starters play a little too long for another set. I'll take another look but at the end of the day, you have to balance between IP, IP/G, SV, and G versus CG from what it sounds like.

[Brian SCEA]

Re: Bullpen tuning in patch 1.04

After the patch, I am seeing Starting Pitchers listed as Set Up pitchers in daily roster mode. In one game the Phillies put in Velasquez as a reliever, and for the Red Sox, Clay Buchholz & David Price are listed as Set Up. (PS3)

Unfortunately I don't know anything about that part of the game, or what changes have or haven't been made there. All I can say is I don't think my changes would affect that.

[bcruise]

Re: Bullpen tuning in patch 1.04

Didn't take me long going through game logs with hook maxed to find something I LOVE seeing post-patch - MANY examples of this:



I can count on one hand the number of times I've seen the CPU pull a starter with a shutout before this change, regardless of where I had the hook slider at - it could be maxed and it still wouldn't pull him. I counted at least 7 - 8 of these up to the all-star break for the Cubs games alone (including those pitchers whom they played against).

That's huge, HUGE for cutting down the number of CG's. Thanks a lot Brian!

[Brian SCEA]

Re: Bullpen tuning in patch 1.04

I can count on one hand the number of times I've seen the CPU pull a starter with a shutout before this change, regardless of where I had the hook slider at - it could be maxed and it still wouldn't pull him. I counted at least 7 - 8 of these up to the all-star break for the Cubs games alone (including those pitchers whom they played against).

That's huge, HUGE for cutting down the number of CG's. Thanks a lot Brian!

I agree, I'd like to work on this more. I'm curious what's important to people: totaled stats, individual game behavior, or situational behavior (division, playoffs, multi-game situations). CG and IP are an example of total stats, "managers pull starters too soon" is individual game, and the closer not getting enough work is a combination of total stats and situational behavior (i.e. the solution requires multi-game thinking).

The problem is, every person I've asked this to will give me a different answer, so it's hard to know whether I'm working on something hard to notice instead of something more important. I'm guessing I can probably work on this two weeks at the most.

I guess that calls for a poll? I'll follow up in a few weeks.

[Bobhead]

Re: Bullpen tuning in patch 1.04

I agree, I'd like to work on this more. I'm curious what's important to people: totaled stats, individual game behavior, or situational behavior (division, playoffs, multi-game situations). CG and IP are an example of total stats, "managers pull starters too soon" is individual game, and the closer not getting enough work is a combination of total stats and situational behavior (i.e. the solution requires multi-game thinking).

The problem is, every person I've asked this to will give me a different answer, so it's hard to know whether I'm working on something hard to notice instead of something more important. I'm guessing I can probably work on this two weeks at the most.

I guess that calls for a poll? I'll follow up in a few weeks.

The CPU doing the best it can to win the game is really #1 here, at least for me. I'd love to see a realistic number of CGs and Shutouts and whatnot, but most importantly, I want the Phillies (who have a terrible bullpen) to leave their dominating starter in the game against me almost all of the time, and I want the Royals (who have a great bullpen at least on paper and from last year's stats) to have a pretty quick hook on their starters as soon as there are signs of trouble.

All in all, the CPU should leave their starter in until the starter is no longer the best option. Counting stats are arbitrary at best, and are secondary in importance to a properly behaving situational AI.

None of this applies in simmed games of course, in that case the priority should be on stats.

Not saying this is or isn't already happening post 1.04... I haven't played a game yet. Just sharing my thoughts. The changes you posted all seem like great/welcome additions.

[SoxFan01605]

Re: Bullpen tuning in patch 1.04

I don't know where else to put this, but these are some changes for the patch. At the time of the patch I had to move on to work on other fixes, so I couldn't clarify these details until now.

* CPU managers will let starters pitch slightly longer in various situations. They're less likely to pull starters over individual runners on base.
* In simulated games, CPU managers will pull starters slightly sooner. I.e. games you sim on the front end and fast forward (both RTTS and manual).
* When a closer hasn't played in 3 or more days, the CPU may make them play regardless of the save situation to give them work.
* As a result of the two previous changes, starters will pitch fewer total IP (and IP/G), while relievers and closers will play more games.
* Other changes to how CPU managers manage relief pitchers in certain situations. These are all tuning changes.

This is excellent information. Thanks for the breakdown and all your work!

(and thanks to bcruise for his testing and great feedback as well!)

[oldtimey]

Re: Bullpen tuning in patch 1.04

You will greatly reduce the number of innings that your middle and setup relievers pitch if you always keep a "6th starter" in the long relief slot in your roster. Your middle and setup guys innings increase when they have to pitch earlier in the game than the 6th and when they pitch multiple innings when a game goes into extras. To keep a long guy in the LRP slot in a 12-man staff, you should only have 1 one SU pitcher. Though there are 2 slots, a LH/RH combo in these slots is not necessary. Just one guy, with stuff that will suffice to occasionally close games, should be in the one slot. You will need all 4 of the MRP slots filled. If you go to a 13-man staff or when rosters expand, fill the LRP slots before the SU slots to further take pressure of your MRP guys. You will find that the CPU now will only use your MRP in the 6th inning or later, keeping their appearances at 1 IP or less. Your relief staff's numbers will then be realistic at the end of the season. You might get relievers with 85 appearances, but you will not get any reliever with close to 100 IP.

[Jr.]

Re: Bullpen tuning in patch 1.04

I agree, I'd like to work on this more. I'm curious what's important to people: totaled stats, individual game behavior, or situational behavior (division, playoffs, multi-game situations). CG and IP are an example of total stats, "managers pull starters too soon" is individual game, and the closer not getting enough work is a combination of total stats and situational behavior (i.e. the solution requires multi-game thinking).

The problem is, every person I've asked this to will give me a different answer, so it's hard to know whether I'm working on something hard to notice instead of something more important. I'm guessing I can probably work on this two weeks at the most.

I guess that calls for a poll? I'll follow up in a few weeks.

I would think that correct situational behavior would alleviate all of these, wouldn't it? A situation in a played game should play out the same way if it were the same situation in a sim, right?

I don't know if this is your area, Brian, but I would also imagine that adding L/R splits for pitchers would help a lot, too.

[SPU32]

Re: Bullpen tuning in patch 1.04

I don't know where else to put this, but these are some changes for the patch. At the time of the patch I had to move on to work on other fixes, so I couldn't clarify these details until now.

* CPU managers will let starters pitch slightly longer in various situations. They're less likely to pull starters over individual runners on base.
* In simulated games, CPU managers will pull starters slightly sooner. I.e. games you sim on the front end and fast forward (both RTTS and manual).
* When a closer hasn't played in 3 or more days, the CPU may make them play regardless of the save situation to give them work.
* As a result of the two previous changes, starters will pitch fewer total IP (and IP/G), while relievers and closers will play more games.
* Other changes to how CPU managers manage relief pitchers in certain situations. These are all tuning changes.

thank you. and is it possible to increase the size of the green-colored, stamina circle when we go to view our bullpen while we are in-game - instead of just before we start the ballgame ?

once again - thank you

oh, .... and could you please fix runners not advancing from second base to third base when there is a ground-out to the right side of the infield, when you are simulating in the fast-forward screen, while playing a game ? they never, ever advance with zero or one out, as they should.

and there is also proof of this in the game logs as well. I have begged for this for like five years and the bug is still there.

thanks

[SPU32]

Re: Bullpen tuning in patch 1.04

And maybe for MLB 17, you could add a slot in PR that says "SIT" for a situational lefty.
It would be great if your sim engine could be akin to OOTP.
Thank you soooo much for caring about Sim Baseball.

[Smallville102001]

Re: Bullpen tuning in patch 1.04

I don't know where else to put this, but these are some changes for the patch. At the time of the patch I had to move on to work on other fixes, so I couldn't clarify these details until now.

* CPU managers will let starters pitch slightly longer in various situations. They're less likely to pull starters over individual runners on base.
* In simulated games, CPU managers will pull starters slightly sooner. I.e. games you sim on the front end and fast forward (both RTTS and manual).
* When a closer hasn't played in 3 or more days, the CPU may make them play regardless of the save situation to give them work.
* As a result of the two previous changes, starters will pitch fewer total IP (and IP/G), while relievers and closers will play more games.
* Other changes to how CPU managers manage relief pitchers in certain situations. These are all tuning changes.

Want few questions I have about this. 1. You said patch number that is already out so was that just missing from the patch notes? Or did you mean this was going to be in the next patch? 2. Will CPU use left righty matchups now? 3. Last thing will CPU pitchers not come out of the bullpen and pitch like 3 innings all the time?

[Smallville102001]

Re: Bullpen tuning in patch 1.04

I'll take a look at CG stats in general next time I have a chance. The stat depends heavily on subjective "borderline" criteria.

The main focus of my changes were on IP/G and the fact that every starter would go an extra half inning per game. That adds up and heavily affects SV and G for relievers and also ends up potentially hiding other logic issues. The original problem was parts of this simming logic is shared for in-game manager decisions, so I had to fix things regardless of if I actually wanted to, in order to address in-game issues.

In other words, in-game logic was pulling starters too soon for one set of reasons, sim logic was letting starters play a little too long for another set. I'll take another look but at the end of the day, you have to balance between IP, IP/G, SV, and G versus CG from what it sounds like.

Sounds like a lot of hard work keep up the great work you guys.

[WaitTilNextYear]

Re: Bullpen tuning in patch 1.04

Aww man, so you're saying I need to re-tune my SP stamina/RP stamina/manager hook sliders? lol, I had them perfect for the old AI. Oh well, thanks for the fixes.

[Smallville102001]

Re: Bullpen tuning in patch 1.04

Hmm...maybe not so fast on that hook/stamina slider combo bombing IP:

Again, starter stamina minimum, hook maximum

Spoiler

That, honestly, is not bad. And I've had a few sims now that look very close to this as far as the leaders go. CG's have been very, very close to looking like real life numbers with this setting. I've probably got a little leeway to even lower hook or raise stamina to get a few more IP, and not throw anything too far out of whack.


Definitely want to hear Tabarnes' thoughts on this since he's the sim guy.

You can get good starting IP with the starter stamina and hook slider before and after the patch but it also causes another problem that I have yet to find a way to fix. That is when you lower the starters IP those IP that they lose are not spread out evenly enough with the bullpen guys so you end up having way to many IP but bullpen guys and way to many games to. I have been running test tonight and ever time I get the starter IP down to a good amount then I am having a lot of bullpen guys with like 100-120 innings and guys pitching in like 130 games. Like I just did a sim with starter stamina at 3 and reliever stamina at 3 and hook at 10 and I had 23 guys with 200 or more IP not a bad number I think in a normal year you have about 30 guys. How ever 32 relief guys threw 80 or more innings with the most being 106.1. Last year in real life the most games innings by a relief pitcher was like 80-82. 27 guys threw 100 or more games out of the bullpen again last year most was only about 80 in real life. So if you get starter IP down then relief pitchers numbers go way way up. So right now it looks like the choice is either have way to many IP by starter or have way to many by relief pitchers. Also I think Closer are getting to many saves. Like ever sim I do I have at least one guy with 57-60 saves and a lot of games with 45-55. This games sim engine for team states and for hitting stats is just about perfect but its pitching stats could use some work. I guess that is not really unexpected though because in game CPU bullpen is like the only thing game play wise that is not just about perfect.