Let's Talk about the manager -- The John Farrell Thread

[radsoxfan]

Given the 3 DPs they had already hit into, I get the bunt idea. 

I'd like to ask Farrell the odds of a coin flip coming up heads after the previous 3 just landed on heads.

More worrisome than the bunt itself (which was dumb enough on its own), is the fact that he may have actually changed his strategy based on double plays earlier in the game.

 

[DennyDoyle'sBoil]

I'd like to ask Farrell the odds of a coin flip coming up heads after the previous 3 just landed on heads.

More worrisome than the bunt itself (which was dumb enough on its own), is the fact that he may have actually changed his strategy based on double plays earlier in the game.

 
Well, in Farrell's defense, baseball games are not coin flips.  When something has happened several times earlier in a game, it's often for a reason, not necessarily chance.  The fact that there were three double plays in six innings can often be because of something that's happening in the game, most usually the way the pitcher is pitching.  Fister's ground ball to fly ball numbers were 13-4.  He was inducing ground balls all game.  Put another way, it's not always a fair coin. 

 

[Kevin Youkulele]

I'd like to ask Farrell the odds of a coin flip coming up heads after the previous 3 just landed on heads.

More worrisome than the bunt itself (which was dumb enough on its own), is the fact that he may have actually changed his strategy based on double plays earlier in the game.

Of course there is a luck component to hitting into DPs, but I think you exaggerate by calling it a coin flip.  Fister has a higher than average GB rate to begin with, and the first 3 DPs plus the fact that only five balls had been hit in the air before Salty's 7th inning at bat (based on reviewing the play by play) suggest that his pitches were particularly heavy/difficult to get under that day.  Would you say that a pitcher who gives up an unusually high number of line drives early in an outing is likely to revert to his career LD-allowed rate for the rest of the outing?  I think not.  The same logic would apply to GBs.

 

[radsoxfan]

Maybe the coin flip isnt the best example since it implies there is absolutely zero weight placed on recent events, but I don't think the strategy changes much because what you've recently seen over a tiny sample size. If Farrell went into the game thinking, "if we're down 2 in the 7th with Salty up against Fister and runners on 1st and 2nd and 0 outs... I'm gonna bunt because Fister is a ground all pitcher" that's one thing. Although I disagree with the logic, at least it shows some consistency in strategy.

But if he just watched the team hit into 3 double plays and then completely changed his strategy because of that recency bias, that's dumb. I put much more stock in overall numbers than 6 innings in one game.

The coin flip analogy wasnt meant to be literal, just imply generally the odds of something happening in the future don't usually change much based on what you've recently watched.

 

[Red Sox Physicist]

 
Hmm. Maybe. Here's the possibilities using linear weights from Spiff's site:
 
[tablegrid= Run Expectancy Matrix ]Outs Bases Exp. Runs 1 out 1,2 0.971 1 out 1,3 1.243 2 outs 3 0.387 [/tablegrid] 
 
I'm a bit burned right now and I can't remember how to use this to figure out what percentage chance he'd have to think he had of turning the double for throwing to second for it to make sense. Is Smas here? He's good at that... 

pdp >= (1.243 - 0.971)/(1.243 - 0.387) = 31.8%
Throwing to 2nd makes sense sense if the probability of a double play was at least 31.8%.

 

[rembrat]

Maybe the coin flip isnt the best example since it implies there is absolutely zero weight placed on recent events, but I don't think the strategy changes much because what you've recently seen over a tiny sample size. If Farrell went into the game thinking, "if we're down 2 in the 7th with Salty up against Fister and runners on 1st and 2nd and 0 outs... I'm gonna bunt because Fister is a ground all pitcher" that's one thing. Although I disagree with the logic, at least it shows some consistency in strategy.

But if he just watched the team hit into 3 double plays and then completely changed his strategy because of that recency bias, that's dumb. I put much more stock in overall numbers than 6 innings in one game.

The coin flip analogy wasnt meant to be literal, just imply generally the odds of something happening in the future don't usually change much based on what you've recently watched.

 
Doug Fister is 4th overall out of ML starters in groundball percentage. John Farrell is better at his job than you are at his job. The facts are that Salty was having a brutal day at the plate, runs like a catcher, and was facing a sinkerballer. I get that some of you HATE the bunt but please don't throw your arms up in the air befuddled at the move.

 

[radsoxfan]

Doug Fister is 4th overall out of ML starters in groundball percentage. John Farrell is better at his job than you are at his job. The facts are that Salty was having a brutal day at the plate, runs like a catcher, and was facing a sinkerballer. I get that some of you HATE the bunt but please don't throw your arms up in the air befuddled at the move.

And he was probably 4th in GB percentage (or close to it) before the game started too. If that's why Farrell called for the bunt, I still disagree with it, but for a different and less vehement reason. If he called for the bunt because he just watched 3 double plays earlier in the game, I agree with his decision even less.

Although given John Farrell is better at his job than we are at his job, it's unclear why this thread exists at all. I suppose we should just accept everything he does as correct and not even debate the merits of any decision.

 

[Red(s)HawksFan]

 
Doug Fister is 4th overall out of ML starters in groundball percentage. John Farrell is better at his job than you are at his job. The facts are that Salty was having a brutal day at the plate, runs like a catcher, and was facing a sinkerballer. I get that some of you HATE the bunt but please don't throw your arms up in the air befuddled at the move.

 
Given the "facts" of the situation, why is asking Salty to bunt supposed to be the unquestioned best option?  This isn't simply a matter of throwing our arms up befuddled at the bunt.  It's befuddlement at having a player who has never demonstrated the ability to bunt being asked to lay one down regardless of whether it was the right move for the situation.  Don't have to be a "hater" of bunting to question or disagree with the decision.
 
Pinch-hit for Salty with a player who doesn't run like a catcher but has actually demonstrated an ability to lay down a bunt successfully.  That solves the two "facts" over which Farrell had any control whatsoever.  Furthermore, he could have sent up a pinch-hitter who, while perhaps still "running like a catcher", might see the ball better and do better than Salty could have by swinging away.
 
While it might be easy to try to break this down into bunt lovers vs bunt haters, it's not that simple.

 

[bosockboy]

Yeah he asked Salty to perform a task he is inherently awful at.  The odds were just as good he runs into one as him executing a perfect bunt.

 

[lexrageorge]

Maybe the coin flip isnt the best example since it implies there is absolutely zero weight placed on recent events, but I don't think the strategy changes much because what you've recently seen over a tiny sample size. If Farrell went into the game thinking, "if we're down 2 in the 7th with Salty up against Fister and runners on 1st and 2nd and 0 outs... I'm gonna bunt because Fister is a ground all pitcher" that's one thing. Although I disagree with the logic, at least it shows some consistency in strategy.

But if he just watched the team hit into 3 double plays and then completely changed his strategy because of that recency bias, that's dumb. I put much more stock in overall numbers than 6 innings in one game.

The coin flip analogy wasnt meant to be literal, just imply generally the odds of something happening in the future don't usually change much based on what you've recently watched.

 
I think the assumption that is flawed is that there was some "change" in strategy, which therefore made the decision worse.  
 
Farrell made a decision to minimize the chances of grounding into a double play.  He likely based that decision on (a) the fact that Salty is a catcher; (b) Fister is a ground ball pitcher; (c) Fister was very much on his game that day; (d) Salty hasn't been hitting well.  There was a cost to that decision:  run expectancy goes down assuming a successful sacrifice, and Salty is a terrible bunter.  FWIW, the last point is something I did not consider in my earlier defense of Farrell's decision, and I do think it should have been given greater weight; it's debatable if Salty's bunt did indeed make the chances of a double play less likely.  
 
In the decision criteria above, (a) and (b) are facts that are unchanged by recent events.  Items (c) and (d) do indeed have recency bias, but sometimes pitchers do get into a groove, and sometimes batters do go into slumps, for reasons that are unclear (or are perhaps known but not revealed, such as a hidden injury).  

 

[radsoxfan]

I think the assumption that is flawed is that there was some "change" in strategy, which therefore made the decision worse.

Well in his post game comments Farrell pointed to the 3 double plays earlier in the game as one of the reasons for the decision. Maybe it was overall a very minor component in the decision but he just happened to bring it up because it sounded good to say at the time.

In a more general sense, I hope Farrell isn't prone to such recency bias as an important criteria in his decision making.

 

[kieckeredinthehead]

Well in his post game comments Farrell pointed to the 3 double plays earlier in the game as one of the reasons for the decision. Maybe it was overall a very minor component in the decision but he just happened to bring it up because it sounded good to say at the time.

In a more general sense, I hope Farrell isn't prone to such recency bias as an important criteria in his decision making.

 
What you are calling "recency bias" is the reason for the in-game manager's existence: to use prior information on the expected outcomes, and temper that with his observations of what is happening on the field that day. The most recent information is often the most relevant. "Gee I hope he doesn't take a few pitches just because the pitcher's thrown eight straight balls!"

 

[radsoxfan]

What you are calling "recency bias" is the reason for the in-game manager's existence: to use prior information on the expected outcomes, and temper that with his observations of what is happening on the field that day. The most recent information is often the most relevant. "Gee I hope he doesn't take a few pitches just because the pitcher's thrown eight straight balls!"

It's not that you have to completely ignore recent events.... but there is a real danger in overvaluing them. In my opinion, if the 3 double plays were a major factor in Farrel's decion, that's placing too much value on those recent events above the more important and stable large sample of background data he has.

This is a problem in all professions. If you ignore the process, and simply react to what you have seen recently, you start chasing your tail and make decisions that are regrettable in retrospect.

Maybe that happened here, maybe it didn't. Only Farrell knows for sure how much weight he placed on the double plays he just saw versus the Salty/Fister match up and game situation itself. I only brought up the point because his statements after the game implied he may have overreacted to a few untimely ground balls.

 

[smastroyin]

 
Doug Fister is 4th overall out of ML starters in groundball percentage. John Farrell is better at his job than you are at his job. The facts are that Salty was having a brutal day at the plate, runs like a catcher, and was facing a sinkerballer. I get that some of you HATE the bunt but please don't throw your arms up in the air befuddled at the move.

 
In 2155 major league plate appearances, Salty has laid down 3 bunts, all in a two month stretch in 2009.
 
In 1922 minor league plate appearances, he laid down 2, the last in 2005.
 
But no, it's not befuddling at all why he would be asked to bunt in 2013.
 
I'm sure he looks great in drills, and I understand there is a tacit assumption that any major league player should be able to bunt.  But, when you combine the low percentages of the bunt actually helping you (it would be one thing if the lead runner represented the tying run) with the low percentage of the guy successfully carrying out the sacrifice, it doesn't seem that smart.

 

[DennyDoyle'sBoil]

It's not that you have to completely ignore recent events.... but there is a real danger in overvaluing them. In my opinion, if the 3 double plays were a major factor in Farrel's decion, that's placing too much value on those recent events above the more important and stable large sample of background data he has.

This is a problem in all professions. If you ignore the process, and simply react to what you have seen recently, you start chasing your tail and make decisions that are regrettable in retrospect.

Maybe that happened here, maybe it didn't. Only Farrell knows for sure how much weight he placed on the double plays he just saw versus the Salty/Fister match up and game situation itself. I only brought up the point because his statements after the game implied he may have overreacted to a few untimely ground balls.

 
What I'm about to say is a bit esoteric (e.g., potentially stupid), and I agree about the point that Farrell is better at his job than I'll ever be.  That said, while I agree managers can overvalue what has happened in the game that day (or the series) and give it recency bias, I actually think that where a manager can be most effective is in seeing when something is happening that contradicts expectations going in.  If a ground ball pitcher is giving up an inordinate number of fly balls, and your batters are coming back to the dugout telling you everything is up, perhaps that's something to pay attention to and not treat the pitcher on that given day as though he were a ground ball pitcher.
 
But, whatever, I'm not even sure what relevance "ground ball pitcher" is to the Salty situation.  On the one hand, there is the obvious -- it makes it more likely that Salty will hit into a double play and thus perhaps counsels toward a bunt.  But isn't one of the reasons you bunt with first and second and no out in a two run game that it dramatically increases your odds to score at least a single run and thus cut it to a one-run game?  I would tend to think ground ball pitchers actually are better at stranding runners at third base with one out than are fly ball pitchers, no?  

 

[Sampo Gida]

Yeah he asked Salty to perform a task he is inherently awful at.  The odds were just as good he runs into one as him executing a perfect bunt.

 
Maybe not, apparently Salty has been experiencing back problems affecting his hitting, hence the day off today.
 
http://www.boston.com/sports/baseball/redsox/extras/extra_bases/2013/09/ellsbury_saltalamacchia_out_with_injuries.html

 

[lexrageorge]

Well in his post game comments Farrell pointed to the 3 double plays earlier in the game as one of the reasons for the decision. Maybe it was overall a very minor component in the decision but he just happened to bring it up because it sounded good to say at the time.

In a more general sense, I hope Farrell isn't prone to such recency bias as an important criteria in his decision making.

Maybe it was just hubris and arrogance, another example of Farrell thinking he knows more about baseball than the writers covering the team. 

 

[AbbyNoho]

Maybe it was just hubris and arrogance, another example of Farrell thinking he knows more about baseball than the writers covering the team. 

 
This is now one of my favorite lines ever. 

 

[Reverend]

Maybe it was just hubris and arrogance, another example of Farrell thinking he knows more about baseball than the writers covering the team. 

 
Wrong forum--Blinded by the Lombartis is to your left.

 

[normstalls]

Masterful job by the manager tonight. The guy has been fantastic this season. Hands down Manager of the Year.

 

[joe dokes]

Well in his post game comments Farrell pointed to the 3 double plays earlier in the game as one of the reasons for the decision. Maybe it was overall a very minor component in the decision but he just happened to bring it up because it sounded good to say at the time.

In a more general sense, I hope Farrell isn't prone to such recency bias as an important criteria in his decision making.

 
What your calling "recency bias" is really the fact that in his post game press conference, Farrell said, "3 double plays" and not, "Fister is one of the best groundball pitchers in the American League who was obviously on his game tonight as evidenced by the number of ground balls our team hit tonight, 3 of which turned into double plays."
 
The bunt was defensible. The identity of the bunter is much, much less so.

 

[Smiling Joe Hesketh]

 
Doug Fister is 4th overall out of ML starters in groundball percentage. John Farrell is better at his job than you are at his job. The facts are that Salty was having a brutal day at the plate, runs like a catcher, and was facing a sinkerballer. I get that some of you HATE the bunt but please don't throw your arms up in the air befuddled at the move.

 
Salty had not laid down a bunt all season long. Fister was tired. The Sox had had trouble scoring runs all day, and yet they gave Fister a free out at the biggest junction of the game.
 
Farrell's a good manager, but the decision was indefensible. Bunting in general is a terrible idea, bunting with a crappy bunter at the plate an even worse one.
 
He hasn't bunted much all season so I'm assuming it was just a bad decision and not a larger pattern, which is fine. He's been overall very good this year, but that game was pretty poorly managed IMO.

 

[radsoxfan]

What your calling "recency bias" is really the fact that in his post game press conference, Farrell said, "3 double plays" and not, "Fister is one of the best groundball pitchers in the American League who was obviously on his game tonight as evidenced by the number of ground balls our team hit tonight, 3 of which turned into double plays."

As I already said, who knows what Farrell was actually thinking. Maybe the recent DPs weren't a big deal and he just threw it out there as a side comment to the media. Quite possible. I disagree with the bunt regardless of the motive, but would be more worried about his thought process if it was overly influenced by a few plays early in the game.

As far as last night, overall I thought he did a good job, particularly with the bullpen. For someone who has seemingly been fond of the pinch runner though, it's confusing why Nava ends up on the bases in crucial spots. He's slow and has awful instincts. No reason for that to be happening post September 1st.

 

[Stitch01]

Im guessing that Salty being banged up factored into the decision, although I think it would have been better to pinch hit.

 

[Reverend]

As I already said, who knows what Farrell was actually thinking. Maybe the recent DPs weren't a big deal and he just threw it out there as a side comment to the media. Quite possible. I disagree with the bunt regardless of the motive, but would be more worried about his thought process if it was overly influenced by a few plays early in the game.

 
I respect the analytic approach, but I think this may be a bit too rigid for thinking in terms of double plays. Double plays, of course, influence the number of plays in the game. Salty bunted in the 7th inning. That means before the 7th inning, at most six double plays were even possible. There were three. And one was in the previous inning and another was one Salty himself had hit into.
 
At some point, pure statistics without introducing actual knowledge of reality falls short; this is the basis of the Bayesian revolution in statistics. It's not accurate to refer to those double plays as just a few plays when we can just as accurately describe them as determinative of the outcomes of half of the previous innings.
 
Having seen Farrell speak on the matter of how he uses analytics, there is no doubt in my mind whatsoever that he knew about Fister's GB tendencies. I'm not saying the bunt was the right decision, but I don't think he approached the decision in a facile manner.

 

[smastroyin]

Just to be pedantic and banal (and probably specious) all at once, I'll point out that through six innings, there could be a lot more than 6 opportunities.  In this case, I believe there were actually 8.
 
For instance:
Lead-off hitter reaches.  First opportunity is with none out.  If the second batter makes an out without making a DP, there is a second opportunity in that inning.
 
Innings where you have multiple hits present many more than 2.

 

[Reverend]

That's true, but I didn't say opportunities, but rather referred to how many DPs were possible. I suppose I could have incorporated third swinging strike balls dropped by the catcher but that didn't seem useful.
 
My only real point was to say that referring to inducing three double plays in six innings as "just three plays" is a bit misleading. And three of eight ain't that bad neither, but I know you're a quant nerd.  (And pretty good at it.) One way or another, Fister was getting it done, and Farrell knew it so he switched things up. Didn't work and I don't love bunting, but I don't think it's incoherent.

 

[smastroyin]

I should have said you were using the wrong denominator.
 
Regardless, as others have said, Salty is the wrong guy for this, even if it was the right call.  For one, if your thought process is that the guy you are sending up to the plate is so gauranteed to not do something positive that you are just trying to have him do less negative, then send a pinch hitter unless there is some reason that you can't.  Beyond the fact that I don't think it was the right call (it's the 7th, you are not advancing the tying or go ahead run to third with one out even with a successful sacrifice) let's look at a little but of quant with made up numbers to give Farrell the benefit of the doubt.
 
Salty Walks or Strikes out in 40% of his PA.  When he puts the ball in play, it's on the ground only 31.7% of the time.  So, 1 in 5 times at the bat, Salty is going to hit a ball that creates a DP opportunity.  Some of those times, he will get a hit, some of those times, the DP will not be turned due to weak contact, an error, etc..  I would call it a 1 in 6 chance at best that Salty hits into the double play.  But, Fister was on and we'll credit old friend Jose Iglesias with increasing the chances of the GB being turned into a DP so we'll just say, again for the benefit of the doubt, 1 in 3 chance for Salty to hit into the DP.
 
Conveniently, perhaps, there is a 33% chance that Salty reaches safely.  Again, since Fister was "on" and Salty was nursing an aching back, we'll call that 20% (this is going to make it so that the correct move was to pinch hit, but whatever).
 
So 33% chance of DP
20% chance of no out and all runners advance at least one base (we can argue about how often the runners would advance more than one base (about half of Salty's hits are for extra bases), but again, we'll skew this towards Farrell's favor)
Giving again the benefit of the doubt, we'll say there was no chance any other out Salty might make would be "productive"
So that leaves a 47% chance of the situation we got, which is to say Salty makes an out without the runners advancing.
 
Last, we can give Salty a generous 75% chance of getting the bunt down successfully (that's league average).  Based on his experience and on the difficulty of properly bunting the sinker working in on his hands, I would actually have put the successful sacrifice % at closer to 50%.
 
I suppose I could do all of the math based on the base-out situations, but even just in my head, when I look at the situation like that, I can't justify a sac bunt.

 

[Bellhorn]

 
At some point, pure statistics without introducing actual knowledge of reality falls short; this is the basis of the Bayesian revolution in statistics. It's not accurate to refer to those double plays as just a few plays when we can just as accurately describe them as determinative of the outcomes of half of the previous innings.
 
Having seen Farrell speak on the matter of how he uses analytics, there is no doubt in my mind whatsoever that he knew about Fister's GB tendencies. I'm not saying the bunt was the right decision, but I don't think he approached the decision in a facile manner.

No idea what you are getting at with the bolded sentence, or what it is supposed to have to do with Bayes.  It's true that a Bayesian approach would be theoretically appropriate here, but it actually makes very little difference to a rational estimate of the DP probability when Salty was at bat in the 7th.
 
And while I think we all know that Farrell is generally sound when it comes to analytics, don't you agree that the fact that he brought up the DPs in response to the question was, at best, an unfortunate slip-up that is worthy of comment in a thread devoted to such discussion?  His comment raises the possibility that this thought process really was in play to some degree, and as radsoxfan has repeatedly pointed out, this is a silly way to go about in-game managing. 
 
It's akin to justifying an IBB by saying "well, he had hit the ball hard against us three times already today."  That's clearly a silly response, and I don't see why the response to the bunting question was any less so.

 

[lexrageorge]

No bunting:
 
33% chance of run expectancy going to 0.357.
20% chance of run expectancy going to 2.232
47% chance of run expectancy going to 0.882
0% chance of any other event occurring (run scoring hit of any kind, productive out, triple play, fielding error)
 
Expected new run expectancy is 0.979, which is well below the 1.42 actual run expectancy with no outs and runners at 1st and 2nd.  This makes sense; we've given Fister a higher chance of a DP, and Salty a lower chance of getting on base than the average player, and have eliminated the 33% possibility of an extra base hit. 
 
Bunting:
 
75% chance of run expectancy going to 1.286.
25% chance of run expectancy going to 0.882
0% chance of other outcome (double play, 1st-and-3rd, or error)
 
Expected new run expectancy if bunting is 1.185
 
So, actually, under these admittedly very contrived conditions, the math says go ahead and bunt.  
 
I did the same for the chances of scoring 1 run:
 
Not bunting:  46% of scoring at least 1 run that inning.
Bunting:  63% of scoring at least 1 run. 
 
So, you can make the math work.  But while I have defended Farrell in this situation, I would never use this math as part of the defense. 

 

[Reverend]

No idea what you are getting at with the bolded sentence, or what it is supposed to have to do with Bayes.  It's true that a Bayesian approach would be theoretically appropriate here, but it actually makes very little difference to a rational estimate of the DP probability when Salty was at bat in the 7th.
 
And while I think we all know that Farrell is generally sound when it comes to analytics, don't you agree that the fact that he brought up the DPs in response to the question was, at best, an unfortunate slip-up that is worthy of comment in a thread devoted to such discussion?  His comment raises the possibility that this thought process really was in play to some degree, and as radsoxfan has repeatedly pointed out, this is a silly way to go about in-game managing. 
 
It's akin to justifying an IBB by saying "well, he had hit the ball hard against us three times already today."  That's clearly a silly response, and I don't see why the response to the bunting question was any less so.

 
I simply meant that you incorporate acquired evidence into your analysis.
 
And no, I don't. I think smas's analysis is pretty much the right approach. I would add two caveats though:

1. The data the team uses is clearly more granular than aggregate numbers while still using methods to retain large-N analysis. As it pertains to this case, I mean that they are looking at how Salty does against certain types of pitchers and how Fister does against certain types of hitters. SkyNet is not operational by any means, but Farrell said that they think in terms of how different hitters hit different pitchers. Obviously, I don't know, but they could have data suggesting concern that Salty had a probability of grounding into a double play there. It's worth reiterating that he had already hit into a double play that game, despite the numbers smas cites--indeed, ironically, thinking he would be less likely to do so again because he already had done so in that game would be a true example of recency bias.
2. Farrell has also indicated that the pitchers like Salty. The pitchers have too. I think that could easily have come into play with the decision to have Salty stay in even though they thought the bunt play was a good idea.

I can see how the 3 DPs reference could be interpreted as recency bias, but as someone said earlier, I think it was just him noting the ground ball tendencies of Fister by way of indicating the examples. I think to assume that he was citing that data independent of him understanding the context is actually a much larger leap.

 

[kieckeredinthehead]

And while I think we all know that Farrell is generally sound when it comes to analytics, don't you agree that the fact that he brought up the DPs in response to the question was, at best, an unfortunate slip-up that is worthy of comment in a thread devoted to such discussion?  His comment raises the possibility that this thought process really was in play to some degree, and as radsoxfan has repeatedly pointed out, this is a silly way to go about in-game managing. 
 
It's akin to justifying an IBB by saying "well, he had hit the ball hard against us three times already today."  That's clearly a silly response, and I don't see why the response to the bunting question was any less so.

 
No, neither are "silly." Imagine you have a relief pitcher. Call him Andrew Miller. He has trouble repeating his mechanics. Some days, he can do it and he's throwing strike after strike. Some days, he can't and he's over the place. You have the ability to quantify this difference before a game. When Miller comes into the game and walks the leadoff hitter, is the manager "silly" to say, "well, he's not really in control, and based on all the previous evidence, he's probably not going to get it back today. Better take him out." 
 
Or take your example. Defending the intentional walk by saying he'd already hit the ball hard against your three times that day would be silly if it was Iglesias, but less silly if it's Miguel Cabrera. Or take the opposite example. Say last night Cabrera came up with first base open rather than the bases loaded. If Lester/Farrell decided to pitch to him because he'd seemed to be off that day (most likely due to the injury), it would be defensible, even though under normal circumstances you'd walk him. "He just didn't seem to be squaring up the ball as well as normal." 
 
"He'd already gotten three double plays today" is short-hand for "Given that Fister is a ground-ball pitcher, and given that he's getting a lot of double plays today, including one to Salty, he's more likely to get a ground ball right now." Do you seriously not get how that's a pretty simple Bayesian framework?

 

[HomeRunBaker]

Playing WMB behind the bag with Iggy leading off the inning? I defend Farrell as much as anyone but really wtf?

 

[Plympton91]

Not a fan of leaving Peavy in the game here.

Edit: leaving Peavy in the game was assinine. Didn't have great stuff, had already struggled through long at bats with Ichiro, and was well over 100 pitches, bullpen is expanded and none of the top guys got used last night.

Ridiculous.

 

[SoxFanInPdx]

I sure hope that Farrell turns into playoff Tito come October. I've had enough of these bizarre decisions.

 

[tims4wins]

What was with the shift, and then pitching Soriano outside?

 

[j44thor]

What was with the shift, and then pitching Soriano outside?

 
What is with the RH shift when you need a DP to get out of the inning?
I just don't get how such a smart manager can make such boneheaded dumb decisions.

 

[meadow11]

How many times can a manager's decisions be second guessed in one half inning?  
 
1.  Leaving Peavy in after 100 pitches.  
2.  Leaving Peavy in after walking the first batter of the inning.  
3.  Allowing Thornton in to face 3 batters (2 lefties but he did not have any command tonight)
4.  Shifting against Soriano when a double play is needed.  
5.  Leaving Tazawa in to face Overbay.  
 
Not saying everyone of these decisions were dead wrong, although none worked out, but I can't remember another time when I'm questioning that many moves in ONE half inning.  

 

[smastroyin]

I think he made some dumb decisions but it's not his fault that Matt Thornton is the worst mid-season pickup since Eric Gagne and should be shuttled back to the DL as soon as possible.  I'll volunteer to kneecap him if it helps.
 
The shift was dumb.  It was a classic case of trying to be the smartest man in the room.  I hate that crap.  Especially since your pitcher then completely failed to execute in a manner consistent with your defensive alignment.

 

[judyb]

I didn't like leaving Peavy in, either, but it's not supposed to be like asking for a miracle to expect a few of your supposedly better relievers to strand a few inherited runners once in a while.

 

[mabrowndog]

 
The shift was dumb.  It was a classic case of trying to be the smartest man in the room.  I hate that crap.

 
My thoughts exactly. Especially when you have a pretty stellar defensive infield in terms of range. You want to shade him to pull? Fine. But leaving the entire right side vacant save Napoli was fucking insane.

 

[OttoC]

I suspect that Peavy could have gotten out of the inning while still maintaining a lead. He might have put a dent in the lead but Thornton had a WHIP of 1.500 coming into this game and Tazawa seems to have already pitched too much this year.
 
On the other hand, Peavy left the game at 117 pitches, so replacing him at the beginning of the inning probably makes more sense than letting him stay in the game.

 

[Kevin Youkulele]

The uniformity of opposition to leaving Peacy in, and other moves in the inning that shall not be named, contrasts starkly with the somewhat balanced argumentation about the Salty bunt from a few days ago.

 

[Stitch01]

The shift plus pitching approach to Soriano simply can't happen again. That was indefensible.

 

[joe dokes]

The shift plus pitching approach to Soriano simply can't happen again. That was indefensible.

 
I think that points to Tazawa's pitching at least as much as the shift itself. I can't believe the goal was to pitch him outside, unless it was supposed to be so far outside that not even Soriano would reach for it.
 
Right now, the only relievers that Farrell trusts are Uehara, Breslow and Workman. You can point to using Workman the night before in the blowout as making him unavailable last night, but I think Farrell wanted to see him go on back-to-back nights (as he will in October), so using him the second night in low leverage made some sense.
 
Leaving Peavy in was indefensible.  That said, doing so should have cost them a couple of runs, not 6. Unless Thornton shows a dramatic reversal, I think the next two weeks will be to determine whether Britton or Morales makes the playoff roster as a second lefty.
 
[not race-based analysis ]Tazawa needs a vacation like the 2 weeks off that Okajima got near the end of the '07 season [/not race based analysis]
http://www.baseball-reference.com/players/gl.cgi?id=okajihi01&t=p&year=2007

 

[Rovin Romine]

How many times can a manager's decisions be second guessed in one half inning?  
 
1.  Leaving Peavy in after 100 pitches.  
2.  Leaving Peavy in after walking the first batter of the inning.  
3.  Allowing Thornton in to face 3 batters (2 lefties but he did not have any command tonight)
4.  Shifting against Soriano when a double play is needed.  
5.  Leaving Tazawa in to face Overbay.  
 
Not saying everyone of these decisions were dead wrong, although none worked out, but I can't remember another time when I'm questioning that many moves in ONE half inning.  

 
Farrell was working with a 5 run lead. 
 
Clearly he wanted to get an extra inning out of the starter, or hoped that Thornton would get it together after a few pitches.  Having three pitchers in a row bungle it - well, it happens. 

 

[Stitch01]

I dont hate having Peavy start the 7th.  Maybe one batter and out would have been better, but he had a five run lead and the bullpen is thin.  I didnt agree with most of the moves that inning, but just thought the one 100% clear fuck up was the Soriano at bat.

 

[Toe Nash]

Wouldn't you want to go easy on a guy you're counting on in the playoffs who has a scary injury history? I don't know how you defend sending Peavy out again. I guess he hadn't worked through three jams yet, just two, so he was good to go.

 

[TFisNEXT]

Wouldn't you want to go easy on a guy you're counting on in the playoffs who has a scary injury history? I don't know how you defend sending Peavy out again. I guess he hadn't worked through three jams yet, just two, so he was good to go.

 
 
I agree with this. There wasn't a good reason to keep riding Peavy after he had thrown over 100 pitches when you have a 5 run lead. At the very least, you give him one batter. When he walked Ichiro, he should have been out.

 

[ivanvamp]

I didn't like leaving Peavy in, either, but it's not supposed to be like asking for a miracle to expect a few of your supposedly better relievers to strand a few inherited runners once in a while.

 
It would have been fine, even, for them to allow Peavy's inherited runners score.  The Sox were up 7-2 at the start of that inning.  They could have afforded to give up a couple of runs there and still been very much in control of the game.