THE BOOK--Playing The Percentages In Baseball

If the runner was guaranteed of scoring the run, the outfielder should have let the ball fall.

Quickest way to think of it is this:
Advancing 1B to 2B, or 2B to 3B, adds 0.17 runs (.017 wins).  An out removes 0.27 runs (.027 wins).  That’s why generally, it’s a bad play.  You’d have to have situations where 0.17 runs is worth say .022 wins, and 0.27 runs for the out is worth .021 wins in order to make the tradeoff.  It happens of course.

For 3B to home plate, that adds 0.14 runs with 0 outs, or 0.34 runs with 1 out.  As you can see, it’s going to be tough to let the ball drop with runner on 3B and 0 outs.  Guy has a great chance of scoring already, so record the out.  But, with 1 out, those 90 feet are very valuable.

The question is really when would you WANT to catch the 2nd out and let the runner score.  That is, when does the 0.34 runs become worth less in wins than the 0.27 runs from the out is in wins?  It would be a reverse small-ball situation.

If it is a “drop” just using run expectancy, then surely it is a 100% drop when the runner is the tying run in the 7th inning.

If any team should know this beforehand, shouldn’t it be the Red Sox with their cadre of Jamesian analysts in the front office?

Sabermetric team, schmabermetric team…

Let’s say the Red Sox have calculated the Win Probability of every voluntary opportunity like this - foul balls you can drop, opportunities for fielder’s indifference, when to hope a bunt rolls foul, etc.  And let’s say they have it on a very easily searchable database… or better yet, it miraculously fits on a handy 3x5” index card.  And let’s say they hire one person who has only ONE JOB in the entire organization, which is to recognize these types of situations as they are happening and look up the win probability of each voluntary outcome.

How fast do you think this person can examine this table and determine the right course of action?

And once he does, how quickly can he relay this information to Drew, who’s running hard straight at a wall, running over the uneven ground of a pitcher’s mound?

I hope you’re being facetious MGL!

I think you can EASILY create some rules of thumb that covers 90%, if not 99%, of runner on 3B situations.

While the poor decision is interesting on its own, I’m equally curious how the Boston reporters will cover this. We all know that the Red Sox are one of the leading users of statistics, so I would think questions about the fielder’s decision are in order. I think it’s the same as if a reporter is covering a rebuilding team and that organization gives a large contract to a mediocre free agent. Why?

In the context of each team, there are questions which deserve to be asked. This is one such case. So far, all I’ve found is a Twitter post from Gordon Edes, an ESPN reporter who covers baseball in Boston:

For those of you who thought Drew shouldn’t have caught that ball, guess what: Tito [Francona] agrees with you

I’d be curious to hear how the question was worded and what the follow-up questions were. I presume Sox coaches understand why the play was a poor one, but I’d like that confirmed. On top of that, was Drew’s decision a simple mistake, a case of poor coaching, some on-field miscommunication, etc.?

Mike, no I was not being facetious at all.  As Tango, says, it is simply a matter of “rules of thumb.”

Just like the manager tells the infield when to play up or back with a runner on third and less than 2 outs, so should he be telling his outfielders when to catch or let drop a long foul fly - in advance of course.  Especially late in a one-run game.

I don’t see any issue with that whatsoever.

I also don’t think that Tito can take any issue with Drew.  How is Drew supposed to know what to do?

The basic rule of thumb is simple:  Catch with 0 outs and drop with 1 out.  From there, you can get more detailed, using the score, inning, count, batters and pitcher.  The most important thing, after the number of outs (and the score of course), is the count.  If it is 3-0 or 2-0, then it might be a catch even with 1 out.  If it is 0-1, or 1-1, and probably 2-1, surely it is a drop.  It might even be a drop with 0-1 and 0 outs.

Drew was quoted on ESPN.com as saying this:

“I had every intention of letting it drop and it was just instinct,” said Drew, who had no chance of throwing out Pena at the plate. “I put the glove out right at the last second and it ended up in there.”

Wow, that does not sound believable to me.  If it was near the foul line, sure.  But it was so far in foul territory, if he had indeed thought about it beforehand, there is no way he catches that.  At least I don’t think so.  I am certainly not 100% sure about that.  I would say there is a 50% chance he is lying…

Not only that, but Scott Atchinson on the mound in a tie game in the 10th of a really important game?

He is almost a replacement level reliever according to my projections…

I agree, you should know it as a rule of thumb. I have seen the ball intentionally allowed to fall at least once in the past 41 seasons of Pirates games.

Here’s a call LaRussa made which lost the game to the Pirates on Tuesday. Game tied 2-2 in the bottom of the 7th, 2 outs, Cedeno at 3rd, Tabata (excellent speed) at 1st, Walker batting. LaRussa decides to have Pujols play off the bag, not hold the runner, who promptly takes 2nd base. Walker singles, driving in 2 (4-2 Pirates). Cards score one in th 9th, lose 4-3.

The Cardinal write up:
Ronny Cedeno led off with a double to center and took third on a sacrifice. McCutchen grounded out and Wainwright walked Tabata, who stole second without a challenge from the Cards.

That brought up Walker, who bounced a single up the middle to bring in two runs and send the Cardinals to defeat for the first time since Friday.

“At this point in the [season], we’ve got to win every game we can,” Wainwright said. “If I even just hold it to one right there, we probably win that game. I think you put it on me, and say I didn’t make good pitches when I needed to. Which is something I pride myself on.”

La Russa stood by the decision not to hold Tabata on first base. He said he was more worried about Walker singling through the hole on the right side than about a steal or a hit up the middle, so he had first baseman Albert Pujols playing off the bag to guard against a single to right.

“You’d have to ask the coaches on that,” Wainwright said. “That’s their call.”

========

For the record, from Gameday (2005-2010), MLB games

Bats Base Outs  GBH   GB  GBH%
   L  101    2 1973 8519  .232 
   L  011    2 1340 6209  .216

To expand the sample size, regardless of outs, left handed batter, runner at 1b, no one at 2nd is .262, all other situations .233.

At best, a .029 gain on a gb, but 53% of the time the batter will hit it in the air.

Not a lot of gain in babip prevention in exchange for letting a potential two run lead into scoring position.

"How is Drew supposed to know?” Because these are “blink” ideas that are drummed into ballplayers from a young age, gone over in spring training and discussed by ballplayers when they get together to talk baseball (and Drew, indeed, did know what to do, he just didn’t do it).  This is a place where “the book” and “The Book” agree.  But come on, ballplayers understand some things without needing to know the math.

I’m sure that Drew had thought about dropping the ball in the seconds he had to decide, and could even imagine how it would be beneficial to the Sox.  But, Breadbaker, this is a situation where you do you need math to be sure and I would guess Drew just took the safe route and made the conventional play.

Right, you have to know the math.  NO ONE would know the right thing to do without the math.  As I said, if the count were 3-0, it might have been correct to make the catch.  They might THINK they know the right thing (after all, they have a 50/50 chance of being correct), but without the math there is not way of knowing if you are correct.

We talk all the time about things a manager or coach might know that would effect a decision that we think is right or wrong and perhaps swing it in the other direction.  This is one of the few things that are 98% ascertainable “on paper.”

So if there is one thing that the front office saber people ought to be telling the manager, it is this…

Here are some interesting rules of thumb from Eric Van on the SOSH website.  They seem reasonable:

The strategy for this play is complex even without factoring in the probability that the offense sends the runner at all. In general, ignoring inning and score, you always take the out if there’s 0 outs or if there is also a man on 2nd. With runners on the corners and 1 out you catch the ball if there’s a 4% chance of throwing out the runner; with just a man on 3rd you catch the ball if there’s a 12% chance of throwing him out.

In this situation (man on 3rd, 1 out), where the runner will certainly score, with an average hitter and pitcher, there’s a handy rule of thumb:

-- Never let the tying run or an insurance run score.
-- If you’re the road team, take the out and let the run score if your lead and the inning add to 9 or more, e.g., with a 4-run lead in the bottom of the 5th or later, a 3-run lead in the bottom of the 6th or later, and a 2-run lead in the bottom of the 7th or later. (It’s possible that you relax the rule very early in the game with a big lead; the published WPA tables only got to a 4-run lead).
-- If you’re the home team, same rule but relaxed a little: in the first 5 innings, use lead + inning is 8 or more.

Van’s rules might be reasonable, but how is a real player going add up the the lead and the innings, etc.?

I haven’t worked out any numbers, but the following seems ‘adoptable’ in real MLB (assuming the numbers bear it out):

Take the out unless it’s the tying or go-ahead run in the late innings…

It doesn’t have to be a coach (or I don’t see a reason for it to be the coach). You could teach 2 players to make those sorts of calls, and have them signal to the outfielders before each at-bat (or mid at-bat) through signaling to the catcher for a brief timeout or somesuch.

Players are constantly signaling to each other anyway. Move in, move out, shift this way, shift that way. There is definitely a lot of coordination going on that someone is obviously in charge of. If it’s hard to do from the dugout, then have the players do it, but there’s no reason to say “It’s too complicated.”

If this was worth .1 runs, and it happens 10 times a season, that’s a run saved which is several hundred thousand dollars, right? And a totally avoidable loss of several hundred thousand dollars at that. Just teach the center fielder and the second baseman (or whoever is visible to a majority of the fielders and has half a brain) the rules of thumb, and get them to signal people, and yell at Drew to let it drop if need be.

If football players can memorize a play book, there’s no reason that a baseball player can’t memorize the rules of thumb above and have it drilled into them to the point it becomes second nature.

It’s not like the score is constantly changing.  When the inning starts, they know what the score is.  So, they are told: “Listen, unless they happen to score 4 runs, if there’s a deep fly and 1 out with a runner on 3B, let it drop in foul”.

I took a lot of graduate classes in sport psychology.  And one thing that stuck out was the fact that people can only effectively consciously remember one thing, when they’re in the midst of doing a bunch of other instinctual things.  Try to remember two or more, and the incremental effect of trying to remember more things ends up being negative.

I guess what I’m saying is, I feel like there are always plenty of things even a MLB RF needs to remember, that are far more valuable over time than this.  I.E. “Pena’s up, so let me shade a bit to the left.” “The wall bends in by the foul line, so if a ball is hit out there I may need to play a carom that bounces sideways.” Etc.

Also, on this play, Drew caught the ball, kind of paused for a moment, then realized he needed to try for home (which actually, now that I think about it, may lend credence to his statement of it being “instinct” to catch it).  The pause and being off balance… if neither of those things were a factor - if the ball was maybe 5 feet closer to fair territory - he may very well have been able to gun down the runner from that spot.

Is it so really clear that the catch is a good idea with a 3-0 count? A walk sets up the DP.

Mike #17, how do things work for football players? I understand that most players don’t really have a complex job, but for the quarterback, wide receivers or running backs, they have to remember their running pattern, who’s going where, etc. That’s a lot to remember too, but they seem to do a fair job (while getting knocked around quite a bit).

Is the type of memorization required to remember running routes/what count to pass on/who to pass to/which side the handoff will be on, etc. somehow different than an outfielder learning positioning, which base to throw to, whether or not to catch a foul ball, etc.?

Is there research about how many items are acceptable, and when decision-making suffers because of the memorization?

What techniques are useful for combating “too many things to remember”? Could we, for instance, make wristbands for each player that show where to position themselves for each batter? Versus LHP on the front side, and reverse it for RHP? Or when they relay outs to each player (holding up the “1” or whatever) they indicate with the other hand which base to throw to, or other info like for pop flies? That sort of confirmation before each play may help, no? Then it’s not one person’s responsibility, but the whole outfields’. Or some sort of coaching to make that second nature based on signals they get.

Or is that our job to figure out—how many different game states need different game signals, and how each player on the field should react?

Coaches tell the fielders where to position themselves all the time. It is trivial for one of the coaches or the manager to signal the corner outfielders to drop a deep foul fly.

It might not be correct at 3-0. I don’t know.  The walk is obviously not bad (for the defense), but the batter is also a much better hitter at 3-1 (after the foul).

It is pretty simple to use the WE tables for this one.  As I always say, if the difference (between the WE for one alternative and the other) is not close, then we don’t care about the particular batter/pitcher matchups, GIDP frequencies, etc. 

If the difference IS close, then we can start to worry about that.

That is a very important point that many of you should remember.  When we do things “on paper” and one alternative is CLEARLY the correct one, then usually those “other things” (that the manager might now, or is not in our model) become inconsequential.  The interesting thing is that if the decision on paper IS close and we need to incorporate other things into “the equation,” then it probably doesn’t matter much what the manager does (because both of the WE’s are probably close to one another anyway), and he can just flip a coin or go with his gut as Tango likes to say…

Sal, that’s a good point.  Though football quarterbacks are actually, in my opinion, quite terrible at doing that.  It’s why Brady and Manning and a couple other guys are successful in the NFL, and guys like Favre, with the exception of last year, truly aren’t.

I guess you could call it “chunking” for the QBs… really, they need to remember just a few things, which they could chunk pretty easily.  “Moss, then Welker, then Faulk” and the rest is kind of instinctual.  I’m not saying outfielders would forget to breathe or run after the ball… I dunno.  Just saying… I think the sample size of plays like this in these guys’ careers is far too small to think they’ve gone through this type of play dozens of times.  My estimate is that this type of play happens twice a year, per team.  Total guess though.

And again… people act like it’s a foregone conclusion that Drew couldn’t get the runner.  If he wasn’t on a BP mound, or the ball was 5 feet closer to the line, or if he just remembered to throw it in there immediately after he caught it, instead of forgetting and being forced into an awkward spin move… well, he has a strong arm and I think he’d double up the runner quite often.  The chance of him doing that really needs to be taken into account when figuring the run expectancy.

@Mike. JD Drew does not have a very good arm anymore.  Before he injured his shoulder he did.  Old reputations die hard.

If he did throw him out it would have been his first assist all year.  He is the only RF’er with more than 30 starts without an assist.

fwiw, the entire first column of the Boston Globe article on the game condemned this play.  Then it said, oh, by the way, the runner was on third only because Buchholz threw away a pickoff throw that wasn’t really even a pickoff throw.

Also, on ESPN last night, Joe Morgan attributed Carl Crawford’s HR last night to the threat of the running game.  He must have said “slide step” 10 times over the remainder of the game.  Oh, also, they implied their running game was responsible for them being 3rd in the league in runs per game despite having a .251 BA (they did mention they led the league in walks, though).  It is a good question, though, TB has slightly above average OBP, slightly below average SLG, good baserunning (but only maybe +12 runs total, less than the gap between them and #4 in the league in RPG), why are they #3?

Interesting - I don’t know much about his stats, but I feel like my eyes see a guy with a decent arm.  (Likewise the fans scouting report sees him at a 3.8-4.0 in “arm” categories, though for a RF that may be below average).  Can’t argue with the stats though… and we’d both agree that it’s not like he has a reputation that keeps runners from taking chances on him.

From what I recall, no player in baseball had more than a zero chance of throwing the runner out (unless he fell I guess)…

Mike #21:  that level of thinking is now sub-varsity high school.  Quarterbacks and receivers at the NFL level are making huge numbers of improvisations as the play starts and keeps going, with route direction, speed and depth revising as the receiver and quarterback see the defensive reactions.  The offensive ‘reactions’ are planned ahead but the decision trees are complex.  There is little mass media public discussion of it.

I don’t understand saying just a few quarterbacks are successful.  What is more amazing is the number that perform well under severe physical duress. Which makes those that still stand out like Manning or Brady all the more amazing. (I disagree on Farve, he has been a top flight QB for 20 years.  I don’t think he needs more to be “successful”.  Considering some of the D he has been ‘provided’, his record is great.  You do get a fair number of ‘what-in-the-world’ plays relative to Brady or Manning or Drew now, but very few match his positives.  I doubt Brady or Manning could have held up the way Farve did last year in his final game--they wouldn’t have been on the field to make that last dumb pass.)

It was a great catch - I was stunned that he got it.  I think HE was surprised he got there, and that surprise could have led to confusion.  What do I know?  I just marvel at the range.  Well, I also marvel at some blaming Drew when it was Buccholtz’s fault the runner was on 3rd to begin with (the pickoff throw was apparently called from the bench, though, which was dumb).

My wife is a Sox fan and when the TV broadcast showed Papelbon and Atchinson warming up side-by-side, I told her that Francona was on the verge of conceding the game (ala Torre in the 2003 WS, putting in Jeff Weaver instead of Mo).  I thought Francona was smarter than the average manager, but woah.  Atchinson?  There/Then??  Indefensible.  He’s their garbage time mop-up man.  How in the world does he get into that game?

I didn’t see the play, but after reading through the comments, I had a few questions.

Should who the hitter is and who is following them in the lineup matter significantly enough to make this situation an automatic drop?

If the situation was an automatic drop, should JD Drew have even risked injury by running for the ball? Should he have just stayed in right field and watched it drop fall?

Similarly, would most fans at the park watching be aware of why he was letting it drop, or would it cause a booing streak/media backlash?

"Should who the hitter is and who is following them in the lineup matter significantly enough to make this situation an automatic drop?”

Apparently you didn’t read too carefully.  Read #20 again.  Of course those things “count.” But typically if it is not close WITHOUT considering those things, then you don’t have to consider them. If it IS close, then you can consider them or flip a coin.

“If the situation was an automatic drop, should JD Drew have even risked injury by running for the ball? Should he have just stayed in right field and watched it drop fall?”

What do you think?

“Similarly, would most fans at the park watching be aware of why he was letting it drop, or would it cause a booing streak/media backlash?”

A. Who cares?

B. Some would be aware and some would not.  There would not be any fan backlash if he dropped it.

Mike #21:  that level of thinking is now sub-varsity high school.  Quarterbacks and receivers at the NFL level are making huge numbers of improvisations as the play starts and keeps going, with route direction, speed and depth revising as the receiver and quarterback see the defensive reactions.  The offensive ‘reactions’ are planned ahead but the decision trees are complex.  There is little mass media public discussion of it.

As a casual fan, this was my impression too. Regardless, unless there is overwhelming proof out there, I would find it really hard to believe that a QB has less to think about than a right fielder. Yet somehow they find a way to do it, so I’d think that there must be a way to do it for right fielders (and baseball in general) better than we are doing it.

It may be a matter of the cost of training a player vs. the reward, but even so, with a run costing hundreds of thousands of dollars, I’d think some team would give it a shot. Maybe teams are giving it a shot and we just don’t see it, I don’t know.

MGL:  but, the problem is what counts as “close”?

seriously.

these sorts of decisions do not naturally come in a binary set.  these sorts of decisions are artificially divided into that set.  there will always be “grey area” decisions.  my larger point, then, is that grouping into “close” and “not close” no longer eliminates a problem: it just relocates that problem.

nick, yes of course, there is no bright line for whether a decision is close or not.  At some point, it is going to be on the border of close and not close.  But, that’s life.  We actually live in an analog and not digital world.  You make the best decision you can, given the data and a cogent analysis of the situation at hand…