A few updates ago I proposed a theme league that died from, well, macromanagement (let's leave it at that).  My proposal was, everybody enter the same team with the same settings, and let's see how Sparky manages it.  Any interest?

I think what you're alluding to above extends even to Mop Up pitchers.  If you have a close game and two pitchers are available, both rested and at 100% -- your ace closer and your mopup -- Sparky will sometimes go with the mopup if the game situation isn't strictly a save situation for which you've checked the "Use in save situations only" box and all other things are equal. 

In other words, if you're halfway through the season and your closer has worked 30 of his 60 innings but your mop has only worked 10 of his 60 innings, Sparky will sometimes bring in the mop over the closer, even in a close game and even if your closer is at 100% and available.  So usage for the entire season factors somewhere into Sparky's decision tree.  Moral of the story is don't leave your mop sitting around if you don't want to see him when you can least afford it.

Please correct me if I'm wrong about this.

As I said before, this is a brilliant thread.

Posted by contrarian23 on 2/22/2011 2:13:00 PM (view original):
One addition I would make here.  I have seen it suggested by several owners that Sparky will occasionally choose a Long B over Long A (or Set Up B over Set Up A, etc) if your "B" pitcher has been significantly underused compared to your "A" pitcher.  I cannot confirm this is true, but (1) it's certainly consistent with how a real life manager would think, especially in long relief situations and (2) since all we have from the Knowledge Base is the statement that "An A designation will almost always be used before a B designation of the same role"  (Italics mine), we have to assume there ARE circumstances where the B will be used first. 

I think this is related to a recent change by admin. From the 12/20/2010 change log:

Improved logic for going with LongB/Setup B before LongA/Setup A when 7th inning and trailing by 5 or more runs.

But even before that, I have seen the "B" guy get used first if it's a lower-leverage situation. Which is fine, and what I want to happen.

Posted by contrarian23 on 7/1/2010 2:12:00 PM (view original):

Algorithm 2: Which relief pitcher should be used?

Remember, this algorithm is triggered by one of three conditions. Either (a) the current pitcher has reached his TPC and there is at least one available pitcher to put in (b) the current pitcher has been pulled for ineffectiveness and there is at least one available pitcher to put in or (c) there is no available pitcher, but the current pitcher has reached his MPC.

Proceeding from situations (a) or (b) is similar, so we can consider them together. We’ll save situation (c) for the next post.

Algorithm 2a (or 2b): Which of the available relief pitchers should be used? I’m going to disregard the use of LH/RH specialists here.

By “available” relief pitcher, we mean a pitcher for whom ALL of the following conditions are true:

• His “relief” box is checked in the advanced settings
• His fatigue is at or above his autorest
• The game is at or later than his “inning available” setting
• If he’s a mop-up, and you have checked “use mop-up only when losing” then your team must be behind
• If he’s a closer and you have checked “use closer in save situations only” then it must be a save situation.

If one or more of the above conditions does not hold, then that pitcher is not considered available, and will not be used if other pitchers are available. For example, in the situation I described above where it’s the 8^th inning and your closer is set to be available in the 9^th inning, if the only other available pitcher is your mop-up, then your mop-up is coming in the game. Regardless of how close the game is, Sparky is going to follow your instructions. You’ve given him 2 pitchers, but for one of them you have explicitly said “don’t use him until the 9^th inning.” Sparky will choose the other guy. One way to avoid this – potentially – is to check the “only use mop-up when losing” box in the manager settings. But this may not solve the problem – now Sparky has to decide which of your two instructions to override, because technically you have no pitcher available (this will be addressed in the next post.)

If you have more than 1 available pitcher, Sparky will choose the one who best matches the game situation. 

• Save situation means your closer is coming in. 
• Late close game means Set Up A if you’re ahead or tied, and either Set Up A or B if you’re behind.  Sparky may possibly use your closer here even if it's not a save situation, provided you have not checked the "save situation only" box.
• Early innings or a late game that’s not that close, you’re going to see Long A or Long B. 
• If the game is reasonably close and none of the above pitchers are available, but you have starters who are allowed to come in as relievers, Sparky will likely use one of them.
• Game out of hand, you’re going to see the mop up.
• If there is no pitcher who correctly matches the game situation, then I believe that roughly speaking Sparky will move down this list until he finds an available pitcher.  So if it's late and close, but your closer and setup guys are not available, Sparky will next look at your Long A, then Long B, then starters, then mop up.

It’s very important to remember here, that once the new pitcher is in the game, Sparky reverts to Algorithm 1 in order to decide whether (and when) to remove him. Once your mop-up is in the game, if you’ve set him to a high TPC (or TPC = none) and pull setting = 1, then Sparky is not likely to pull him, even if the game suddenly gets close.

 

"Early innings or late in a game that's not close, you're going to see Long A or Long B."

I wish this was true. Here's my scnario:

1st game of season (no one fatigued)...winning by 7 runs heading into bottom of 9th. Nine (count 'em, 9!) man bullpen.

Not a save situation, so closer's not available. Not losing, so mop up is not available.

3 Setup A available
3 Setup B available
1 Long A available

Sparky puts Jack McDowell (setup A) in to work a meaningless 9th.

Interesting, Bribar.  Food for thought that your Mop would be the pitcher of choice in all the scenarios above with a 6-run LEAD or greater.  I usually set my Mop to unlimited TPC/MPC and a Pull Setting of 3.  With a Pull of 3 and no PC limits, he could squander an awful lot of a 6-run lead (as in all of it) before Sparky sees that something is wrong.  I've seen that happen several times, and I'm sure it's a major contributing factor to 13-12 and 14-13 games in the SIM -- 13-6 blowouts that suddenly became 13-12 or worse still, 14-13 heading in the other direction.  I wonder how well Sparky assesses the need for a different pitcher on a batter-by-batter basis "in reverse" in those 6 outs or less scenarios listed above -- e.g., as a lead is rapidly shrinking.

Bumped for mrk14471

Posted by teaparty on 3/2/2011 4:13:00 AM (view original):

Posted by contrarian23 on 7/1/2010 2:12:00 PM (view original):

Algorithm 2: Which relief pitcher should be used?

Remember, this algorithm is triggered by one of three conditions. Either (a) the current pitcher has reached his TPC and there is at least one available pitcher to put in (b) the current pitcher has been pulled for ineffectiveness and there is at least one available pitcher to put in or (c) there is no available pitcher, but the current pitcher has reached his MPC.

Proceeding from situations (a) or (b) is similar, so we can consider them together. We’ll save situation (c) for the next post.

Algorithm 2a (or 2b): Which of the available relief pitchers should be used? I’m going to disregard the use of LH/RH specialists here.

By “available” relief pitcher, we mean a pitcher for whom ALL of the following conditions are true:

• His “relief” box is checked in the advanced settings
• His fatigue is at or above his autorest
• The game is at or later than his “inning available” setting
• If he’s a mop-up, and you have checked “use mop-up only when losing” then your team must be behind
• If he’s a closer and you have checked “use closer in save situations only” then it must be a save situation.

If one or more of the above conditions does not hold, then that pitcher is not considered available, and will not be used if other pitchers are available. For example, in the situation I described above where it’s the 8^th inning and your closer is set to be available in the 9^th inning, if the only other available pitcher is your mop-up, then your mop-up is coming in the game. Regardless of how close the game is, Sparky is going to follow your instructions. You’ve given him 2 pitchers, but for one of them you have explicitly said “don’t use him until the 9^th inning.” Sparky will choose the other guy. One way to avoid this – potentially – is to check the “only use mop-up when losing” box in the manager settings. But this may not solve the problem – now Sparky has to decide which of your two instructions to override, because technically you have no pitcher available (this will be addressed in the next post.)

If you have more than 1 available pitcher, Sparky will choose the one who best matches the game situation. 

• Save situation means your closer is coming in. 
• Late close game means Set Up A if you’re ahead or tied, and either Set Up A or B if you’re behind.  Sparky may possibly use your closer here even if it's not a save situation, provided you have not checked the "save situation only" box.
• Early innings or a late game that’s not that close, you’re going to see Long A or Long B. 
• If the game is reasonably close and none of the above pitchers are available, but you have starters who are allowed to come in as relievers, Sparky will likely use one of them.
• Game out of hand, you’re going to see the mop up.
• If there is no pitcher who correctly matches the game situation, then I believe that roughly speaking Sparky will move down this list until he finds an available pitcher.  So if it's late and close, but your closer and setup guys are not available, Sparky will next look at your Long A, then Long B, then starters, then mop up.

It’s very important to remember here, that once the new pitcher is in the game, Sparky reverts to Algorithm 1 in order to decide whether (and when) to remove him. Once your mop-up is in the game, if you’ve set him to a high TPC (or TPC = none) and pull setting = 1, then Sparky is not likely to pull him, even if the game suddenly gets close.

 

"Early innings or late in a game that's not close, you're going to see Long A or Long B."

I wish this was true. Here's my scnario:

1st game of season (no one fatigued)...winning by 7 runs heading into bottom of 9th. Nine (count 'em, 9!) man bullpen.

Not a save situation, so closer's not available. Not losing, so mop up is not available.

3 Setup A available
3 Setup B available
1 Long A available

Sparky puts Jack McDowell (setup A) in to work a meaningless 9th.

This scenario is exactly why I went looking for a thread like this.  This happens to me ALL THE TIME, and it's currently the #1 frustration that I have with the sim.

Posted by contrarian23 on 3/8/2010 1:15:00 PM (view original):
For those interested in reading more about this topic, here are some good threads worth checking out in the knowledge base:

http://www.whatifsports.com/knowledgebase/KnowledgeBase.aspx?pid=0&tid=-1

Article 15: How Sparky chooses when you have multiple available pitchers with the same role designation.
Article 30: How the mop-up is used.
Article 36: What the different roles mean and some general guidelines on how Sparky uses them
Article 39: How autorest works (key point: a pitcher fatigued below his autorest setting is treated exactly as though you set him to "rest")
Article 41: TPC vs MPC
Article 46: When Sparky will use a pitcher who had been set to "rest"
Article 47: Why a pitcher may be in the game (this article heavily influenced my outline of "algorithm 1")
Article 770: How LH/RH specialists are used
Article 591: How the SIM determines its pitching role recommendations (one read of this will make you realize you should never just take these as given)

Thanks for this.  Article 36 was especially helpful.  I've been playing this game for a long, long time, and it never had occurred to me that Sparky will only use a long reliever early in the game (unless all other RPs are tired/unavailable).  I still don't like it, but at least now I understand it.

We need a setting for setup men that's like the setting for defensive replacements:  

• Use setup men in the X inning or later if ahead by fewer than Y runs
• Use setup men in the X inning or later if trailing by fewer than Y runs

That way you could set it so that your Setup A doesn't enter the game in the 9th inning when you're up by 6 or 7 runs.

We already have settings that determine the usage for starting pitchers, closers, and mop-ups.  With one more set of settings we could determine when Sparky uses setup vs. long, and we'd have just about every category covered.

Posted by rasluggo on 3/7/2010 11:35:00 AM (view original):
With which I largely agree, contrarian23.

Because I find this an interesting and worthwhile thread, I would make a couple of more general observations, putting me up to about 4 cents worth.

Any owner is smarter than Sparky, because owners are human and Sparky is not. Nevertheless, Sparky is in one way---his ability to compute---superior to most humans. Humans cannot see, or do not care about, what Sparky sees; and Sparky certainly doesn't care about what humans see.

Secondly, let's look at the question of "caring." Humans do care, at least sometimes, whether a certain pitcher is brought or not, and such is the basis of the complaints. Recently, across many disciplines, there has been a burst of literature hashing over the not-unfamiliar proposition that humans are risk-averse when they make decisions. Most of them, some contend, would rather avoid losing something than taking a risk to gain something.

I suggest, then, that human baseball mangers quite often will manage in a "risk-averse" fashion, but Sparky never will; he will only manage so as to maximize the overall probability of gain. By and large, as you point out, this tends to work. Consider, however, the special situations, which while rare can nonetheless be spectacular.

By the latest reports, the Air France flight went down last year because, being strictly faithful to available inputs and to the absence of key inputs due to icing, the fly-by-wire systems and computers on the plane abruptly shut themselves down, the effective result, which was complicated, being to deny control of the aircraft to the men flying it. In another case, it has been credibly suggested that one of the factors in the financial meltdown---and in no way am I suggesting that human wrongdoing was not everywhere present in the matter---was that too much trust was placed in certain computer alogorithms. These algorithms worked fine, most of the time---except that they relied 100 percent on a certain slice of historical data, which did not enable to them to understand or react when something unexpected happened.

The point being, with regard to pitchers and baseball games, that 10 games a season blown because of what most humans would see as "incorrect" decisions on Sparky's part, are a lot of games to lose in that way, in highly competitive leagues.

(Humans being humans, also, there is surely a bias in all the complaints in the direction of looking at how many times Sparky gets it "wrong" rather than how many times he gets it "right.")

Still, I would like to see Sparky be much more flexible, i.e., be more willing to relinquish control to human managers.

Undoubtedly, as you say, this is quite difficult indeed for programmers to accomplish; and in a business sense, you get the programming you are able to pay for. That the sim works as well as it does is a credit to its creators.

Good stuff.  I think the stuff about risk-averseness is especially interesting.  However, I would argue that Sparky is too risk-averse, at least in the case of my aforementioned complaint about setup A being used instead of long A late in games that aren't all that close.  

I think the disconnect is that Sparky manages every game independently, like it's the 7th game of the World Series.  The reason that the usage of Long A is preferable to Setup A (to me, anyway) when up by 7 runs in the 9th inning is that you only can use your setup A in a limited number of games per season, and you'd rather save him for more high-leverage situations.  But Sparky doesn't see that.  Sparky says "I can't use the closer, I can't use the mop-up (because of settings), my best chance of winning the game is therefore to use the setup A."  I don't have a problem with this; I just wish we had setup usage settings as I mentioned above.

Posted by Bribar on 10/9/2011 3:15:00 PM (view original):

Here's a ticket response I received 2 years ago that may be of some help to somebody: 17 QUESTIONS REGARDING LAST 6 OUTS: I've been here about 8 years and I still don't know what the pitching settings mean, and I think I'm wasting good pitchers at the end of games. If I have 6 pitchers, all at 100% and available for 20/30 pitches in the 8th, who comes in if the score is tied (assume the closer and mopup boxes aren't checked): Closer/Setup A/Setup B/Long A/Long B/Mopup? Same question (8th inning, need 6 outs) with a 1-run lead. Same question (8th inning, need 6 outs) with a 2-run lead. Same question (8th inning, need 6 outs) with a 3-run lead. Same question (8th inning, need 6 outs) with a 4-run lead. Same question (8th inning, need 6 outs) with a 5-run lead. Same question (8th inning, need 6 outs) with a 6-run lead. Same question (8th inning, need 6 outs) with a 7-run lead. Same question (8th inning, need 6 outs) with an 8-run lead. Same question (8th inning, need 6 outs) with a 1-run deficit. Same question (8th inning, need 6 outs) with a 2-run deficit. Same question (8th inning, need 6 outs) with a 3-run deficit. Same question (8th inning, need 6 outs) with a 4-run deficit. Same question (8th inning, need 6 outs) with a 5-run deficit. Same question (8th inning, need 6 outs) with a 6-run deficit. Same question (8th inning, need 6 outs) with a 7-run deficit. Same question (8th inning, need 6 outs) with an 8-run deficit.
6/26/2009 11:37 AM	Customer Support
One piece can fluctuate and that's whether the game is a blowout or not which is based on score and ballpark. the more offensive the park, the bigger score differential needed for a blow out. We'll answer this assuming a neutral park. This is all for SLB. HBD is slightly different, but not much. Assuming closer can enter in 8th inning based on Min Inning Available setting and bases are empty. Same question (8th inning, need 6 outs) with a 1-run lead. Closer A Same question (8th inning, need 6 outs) with a 2-run lead. Closer A Same question (8th inning, need 6 outs) with a 3-run lead. Closer A Same question (8th inning, need 6 outs) with a 4-run lead. Setup A Same question (8th inning, need 6 outs) with a 5-run lead. Setup A Same question (8th inning, need 6 outs) with a 6-run lead. Mopup Same question (8th inning, need 6 outs) with a 7-run lead. Mopup Same question (8th inning, need 6 outs) with an 8-run lead. Mopup Same question (8th inning, need 6 outs) with a 1-run deficit. Setup A Same question (8th inning, need 6 outs) with a 2-run deficit. Setup A Same question (8th inning, need 6 outs) with a 3-run deficit. Setup A Same question (8th inning, need 6 outs) with a 4-run deficit. Setup A Same question (8th inning, need 6 outs) with a 5-run deficit. Long B Same question (8th inning, need 6 outs) with a 6-run deficit. Mopup Same question (8th inning, need 6 outs) with a 7-run deficit. Mopup Same question (8th inning, need 6 outs) with an 8-run deficit. Mopup

Wow, this says it all.  The fact that Sparky goes from using setup with a 5 run lead to using mop-up with a 6 run lead demonstrates how unwilling he is to use long relievers late in games.  I think any reasonable person would agree that long relievers should appear somewhere in the middle of that continuum.

I suppose I could get my desired results by setting all of my non-closers and non-setup men to mop-up, and un-checking the mop-up box.  However, I'd then lose the functionality of being able to use my worst pitcher in the biggest blowouts, but that trade-off might just be worth it.