THE BOOK--Playing The Percentages In Baseball

I would argue that defensive indifference and an intentional walk are NOT the same thing.  An intention walk is largely a function of the pitcher (or manager) wanting to avoid you and pitch to the person behind you.  And the reason they want to avoid you is because of your ability as a hitter.  Good hitters get intentionally walked more than average ones.  Defensive indifference is more equal opportunity.  It is more a function of the game situation, and who is hitting (not who running).  I would think that good base stealers advance on DI about as much as average ones do.

Well, they can never be EXACTLY the same thing, but they are in the same ballpark: the defense doesn’t want to deal with you. 

“I would think that good base stealers advance on DI about as much as average ones do. “

I would bet that the distribution is NOT random.  For example, Table 133 of The Book shows that per 1280 runners on first base, there were 3 DI with a LHP and 6 with a RHP.  This is based on 4 years of data.  The difference is more than 3 SD from the mean.

Table 134 and 135: DI is twice as likely to be called with a lefty batter than a righty batter.

All told, the combination of lefty pitcher, righty batter has 2 DI per 1280 opps.  The righty pitcher v lefty batter has 7 DI per 1280 opps.

I don’t think it’s a stretch to say that if the handedness of the batter and pitcher matter in whether a DI play is called, then certainly the speed of the baserunner is also going to bias the data.

Here’s a hint: if a human being is involved in recording data that requires his judgement, it’s a guarantee that the data will be biased in some form or other.

If we have the opportunity to have unbiased data, we should take it.  An example would be WP and PB.  A knuckleballer is involved in a disproportionate number of each.  If a PB was truly the catcher’s responsibility, then the human scorer would need to take into account that the catcher is trying to catch a knuckleball.

Therefore, while I have no problem with the recording of WP and PB, I have a huge problem is assigning WP to the pitcher and PB to the catcher.

In reality, the category should be “Uncaught pitch” (UP).  And that is subdivided into WP and PB.  The UP would be tracked for both pitcher and catcher.  If you ALSO want to include a separate WP for the pitcher and a separate PB for the catcher, that’s fine.  But, you start with UP.  You do not show WP with PB or without UP.  And same with PB.

This goes directly to all the baserunner scoring notations, with SB, CS, BK, PK, WP, PB, DI, and anything else I may be missing.  You need to AT LEAST show:
runner-safe (RS) and runner-out (RO)

From there, show the bigger breakdown.  But, to rely on the scorer so that we only see a portion of the events is beyond silly.  Why in the world would we, as observers, want data discarded so easily?

The scorer serves a great function, in that he notes his observations.  But, those observations does not supercede he events; it complements them.

DI: Correct.
WP and PB: Correct.
IBB:  I have no freakin’ idea, nor do I care, since as Tango says, as long as these things are recorded separately, an analyst can do with them what he wants, to suit his needs.

I’d argue that they are very different.  An IBB occurs most often in crucial situations where the hitter is very important, and the defense/pitcher is “scared”. 

Defensive indifference occurs when the runner is completely unimportant. 

I still think it should be tracked, I guess, but a good hitter “earns” his IBB by being good.  A runner only gets to 2nd on defensive indifference, because, like the name says, the defense doesn’t care.

"Not really. Hirdt noted that Rule 10.07(h) states that a runner cannot be nabbed with a caught stealing if he would not have been credited with a steal if he had been safe.”

Wow, didn’t know that.  I’d love to see the argument about that one…

#6, what argument?  That is 100% logical.  It is simply a non-entity from the standpoint of the individual players (at least the runner), as it should be, although I don’t know if the catcher gets a CS or an assist on that.  Probably not a CS, but he may get an assist and the fielder may get a putout.  Hey, there are no perfect solutions to these things, as much as we would like to think there are.  When I say, “Correct,” I am being somewhat facetious, as I think that declaring these things “correct” or “silly” is actually silly…

Yeah, after I wrote it, I meant inconsistent.

Jon/5: as I said, if something is not exactly the same, you are going to point out the differences, and I’m going to point out the similarities.

You are also presuming that the name of the category is how the scorer is judging it, when I have already shown evidence that it is biased based on the batter and pitcher handedness, and I have no doubt that it will also be biased on the runner’s speed.

#7 - I mean the argument that would take place in a game in which it occurred.  Maybe it wouldn’t be too intense since the team that wouldn’t get the out would presumably have a big lead, but still…

Silly: you don’t get a SB credit for it.

Correct: you do get a BB credit for pitching indifference (IBB).

I have come to the exact opposite conclusion.  It is silly to count IBB in with non-IBB walks.  They have vastly different run values and even more different values on the chance of winning the game.  The IBB doesn’t have anything to do with a batter’s skill of strike zone determination or pitch identification or swing control, where a batter uses all three of these skills to earn a non-IBB walk.  IBB and NIBB should be kept as completely separate categories.

You are also presuming that the name of the category is how the scorer is judging it, when I have already shown evidence that it is biased based on the batter and pitcher handedness, and I have no doubt that it will also be biased on the runner’s speed.

You haven’t shown any evidence of scorer’s bias that I have seen.  That DI happens more often with right handed pitchers and left handed hitters is completely logical and unbiased if the primary criteris for DI is whether a runner is being closely held at first by the first baseman or not.  DI happens when the run represented by the runner on first has little or no value to the chances of the fielding team of losing the game.  But the fielding team still has to weigh the relative out value of having that runner on first rather than on second compared to the relative out value of having the first baseman play back rather than up.  With a right handed batter you lose almost no out value on the batter by having the first baseman play up and you gain out value on the runner.  With a left handed batter the opposite is true. 

From 2005-2008 there were 940 DI according to Retrosheet.  Ichiro led all runners with 16.  The run value of a DI was .036 runs.  A DI was awarded 22 times when the team at bat was ahead, 61 times when the score was tied, and 854 times when the offensive team was behind by 2 or more runs and 3 times when they were behind by 1 run(?).  18 times out of the 854 the batting team came back to win the game, a rate of .021.  This almost always occurred as the result of a later HR which would have scored the runner with or without the DI.

I don’t see the WP-PB decision to be ripe for scorer bias either.  A WP is almost always given if the pitch hit the ground before reaching the catcher’s mitt, and a PB if it hit the mitt first, unless the catcher had to move significantly from his normal catching position.  Scorers almost always get this call correct.  MLBAM Gameday seems to be labeling both balls and strikes that are in the dirt.  I don’t know how consistent they are with that information.  But if they are consistent, it should help to determine a catcher’s value in preventing UP.

Gotta agree with pretty much everything that Peter said…

If Ichiro is getting 1.7% of all DI, which is probably 4 times the normal rate for a guy of his times on base, then clearly we have a bias.

I know what the win value of a DI play is (it’s in The Book).  That doesn’t take away from the fact that it’s given out in a biased manner.

All I’m saying is to count the DI on a player-by-player basis.  That Ichiro leads the league in DI and that Derek Jeter is the career leader in reaching on error makes it pretty clear that we don’t expect these things to be random.

If nothing else, since the reason for DI is usually that it has very little win value, we really shouldn’t care too much about it, even if some players have more than others for some non-random reason.  Not so for an ROE of course.  Two completely different things.

#2 on handedness - wouldn’t a team in a DI situation be more likely to not hold a runner on first with a left handed batter so the 1st baseman can play deeper?  Usually the strategy is to just get the hitter out and not throw on a steal attempt to avoid errors.  So it would make sense as to the strategy and if you want to call it a bias you can.

In the same vein, wouldn’t guys who get on base a lot at the top of the order get more DIs since they get more PAs, get on base more and have dangerous hitters behind them where it would be advatageous to not hold a runner?

My OPINION is that DI, ROE and CI happen so infrequently that it is interesting to look at, but really adds no information on the ultimate goal of scoring a run when looked at over large samples.  If you wanted to account for each base advanced, then you would need to include these and other types of events.

If we want to credit runners with an SB on DI, we should also go back in time and give SB credit for advancing two bases on a single or three on a double like in the old days.  This way, every base advanced receives credit.

Tango - I am not sure what your point is in post #12.  Does Ichiro advance to 2nd more often than an average runner in DI situations?  Yes, of course he does.  Seven times as often.  Does that indicate a bias that needs to be corrected? No, not at all.  Give him the correct run value for advancing to 2nd in a DI situation of .036 runs ( not the incorrect inflated .12 runs that you have in the book).

13 of 16 of Ichiro’s DIs occured with 2 outs in the ninth or later inning trailing by 2 or more runs with the number 2,3, or 4 hitter batting.  He was also credited with 1 SB in that situation.  Is there any scorer bias? No, the scorers determined that 244 of the 256 advancements in that situation were DIs and 12 were SBs.  Ichiro ran 56% (14 of 25) of the time that he was on first in that situation.  The average runner was less than 10% (257 of 3141).  Does that mean that Ichiro should be credited with a SB instead of a DI?  Of course not!  Not if the defense made no attempt to keep him from advancing.  The situation dictates whether the advance is a SB or a DI, not the frequency that it occurs.  No throw to second, no holding the runner on first plus a situation where a single run doesn’t change the outcome of the game equals no SB.  IT’s that simple.

The point of the bias is that players may be given SB when they should get DI, and others are getting DI when they should get SB.  Well, probably not the last part, but certainly the first part.

So, while Ichiro got 16 DI, while the rest of the league got 924, perhaps the rest of the league should have received 1924 or 2924 DI.

And perhaps the handedness situation hands out DI in a disproportionate number even more than they are currently handing out.

The point is that there is probably a bias, and someone is going to be affected disproportionately.

And we should not act as if the DI are being handed out properly.  A DI may have a .04 run value and a SB may be .18, but the way they are handed out, perhaps an Ichiro SB is .19 and the others are .17, simply because of DI/SB misclasifications due to bias.

Tango - Perhaps you should read my last paragraph in post 15 again carefully.  The scorer is calling 95% of all advancements to second DIs in that situation.  He called 93% of Ischiro’s advancements DIs.  That situation accounted for 81% of all of Ichiro’s DIs.  WHERE IS THE EVIDENCE OF BIAS?

First note that the win probability does increase by moving the runner out of a force position.  It is not completely neutral (though it’s pretty close to neutral).

You are right, I didn’t read the second paragraph (at all actually… reminder to me to not go on this site and play with my kid at the same time).

Ok, so of the 256, 12 are called SB, so of the 14 advancement plays for Ichiro, 0.7 should be called SB, and 1 was.  No bias.

As I also said, do NOT count it as a SB.  I said the way to count it is to call it “runner safe” and “runner out”, as well as doing a further breakdown to SB, BK, WP, DI, etc, etc, etc.

So, I have no idea what we are disagreeing about.  I’m saying to record everything, and not to throw any data out.  That the data recorder should record data, and the data analyst should analyze the data.  Indeed, that Peter can generate those great splits is something that we should do, and since MLB doesn’t have DI as an official category, we’d have never known it without Retrosheet.

I was also suggesting that there is bias based on runner speed.  I may be wrong there.

So, if someone want to explain what my position is that you are disagreeing with, please do so.  Because, as it stands, I don’t see anything we are disagreeing with (other than runner speed possibly being a bias).

In my post #10 I put your positions that I was disagreeing with in italics.  Until your last post it didn’t seem to me that you had changed your positions that I had quoted at all.

I think everyone agrees that we should record everything.

But I would quibble a bit with the spirit of Peter’s posts, which suggest there is essentially no hitter skill reflected in cases where the defense makes little or no effort to stop an advancement.  Clearly, in these DI situations the speed of the runner does matter.  If Adam Dunn tried to advance, even if he’s not being held at 1B, the chances of getting an out are high enough that the catcher likely makes a throw.  By advancing, Ichiro is helping his team (though only a very small amount).  This is even more clear in the case of the IBB, which is heavily influenced by the talent of the hitter—I don’t know what Peter means when he says “The IBB doesn’t have anything to do with a batter’s skill....” IBBs of course do have real offensive value, and those issued to the hitters who receive them most often are likely more valuable than average. 

There are other cases where the defense makes less effort to prevent an advance, because they have other priorities.  With a 3 run lead in the 9th, an outfielder may not throw home on a runner scoring from 2nd base on a single—even though he has a good chance of recording an out—because it would allow the hitter to advance to second.  Should we debit the baserunner 1 run scored?  For that matter, if a Seattle hitter gets a single in the 1B/2B hole because Ichiro is being held on 1B, should we credit him with only .9 singles since Ichiro really deserves some of the credit? After all, the 1Bman was not making his best effort to get the hitter out.

There’s a continuum here, and the defense is often weighing multiple factors in determining its priorities.  In the case of DI, the win value of the advancement is very low, and it certainly makes sense to account for it separately.  And if you had to choose between ignoring it and treating it as a SB, you would do the former.  But given the same choice on IBBs, it’s much better to treat it as a BB than, well, an out or whatever else the alternative might be.  And the talent of the baserunner/hitter impacts all of these to some extent—I doubt it’s truly random for any of these plays.

Peter, ok in the post 10 you are quoting, you just quoted the part where I said it was silly to not give a SB for a DI play, while at the same time giving a BB for a IBB.  It was inconsistent.  I further went on to say how I would handle the scoring of SB and DI play (have an umbrella of runner-safe, runner-out, which is further subdivided).

So, if we disagree, it’s in my analogy.  There’s no further disagreement on that part.

***

The question of bias is that while there is a disproportionate number of DI being given out based on batter and pitcher handedness, the argument is that we expect the disproportionateness (that a word?  it is now!) because the 1B is not holding on the runner because of those handedness.  Therefore, the question is how disproportionate should we see the numbers.

I agree I have not shown the bias to that level of context, nor based on runner speed.

***

With that all said, is it really necessary to get the scorer to tell us?  Based on the win expectancy charts, we know the potential win impact of the plays.  The scorer is doing among other things:
1. looks at the fielder alignment
2. considers the inning/score

According to 10.07(g):
http://mlb.mlb.com/mlb/official_info/official_rules/official_scorer_10.jsp

Rule 10.07(g) Comment: The scorer shall consider, in judging whether the defensive team has been indifferent to a runner’s advance, the totality of the circumstances, including the inning and score of the game, whether the defensive team had held the runner on base, whether the pitcher had made any pickoff attempts on that runner before the runner’s advance, whether the fielder ordinarily expected to cover the base to which the runner advanced made a move to cover such base, whether the defensive team had a legitimate strategic motive to not contest the runner’s advance or whether the defensive team might be trying impermissibly to deny the runner credit for a stolen base.

But, I don’t need him to consider the inning, score.  I can do that myself.  It’s objective information.  What he CAN do is tell me that the 1B is not holding the runner.  He can tell me that the middle infielders are not covering secondbase.  He can include that as an additional notation.  He can tell me that the catcher did not even set himself up to even attempt at positioning for a throw.

He can call that “indifferent” (or quatloo).  But, I don’t need him to consider the score, because the analyst can do that himself.

As it stands, I am sure there are plenty of SB being given out that should be DI.  And, I have no doubt that the number of DI given out era-to-era has changed drastically.

***

Again, my overriding point is to record what you observe, and not try to make an interpretation to such an extent that you create potential bias.

I agree with Tango that there doesn’t really seem to be any disagreement here.

In the dissenting comments, I wonder if the underlying discomfort is around the way we assign credit to players.  Right now, someone who gets 30 stolen bases is considered a very good base stealer.  The worry may be that giving people SBs for DI will mutate those long held valuations of speed.  If someone happens to get 10 DIs in a season, that will warp his SB total beyond recognition.

To me, that’s really beside the point.  If analysts can’t change their, “Wow, Reyes got over 80 SBs last year,” to “Wow, Reyes got 80 SBs, not including DIs,” than they shouldn’t be getting paid to analyze baseball.

Regarding IBBs, sure some are based on the skill of the batter, but many are based on the situation and not the batter.  And whether it’s skill or circumstance, they are used the same way as regular walks when calculating a whole host of stats many analysts depend on.  So I don’t really see the difference between an IBB and a DI.  Some are the skill of the player, and some are the circumstance.

Anything that shines further light is good, so eliminating as many judgement calls as possible from the score keepers and maximizing the granularity of the data from the score keepers is a worthwhile endeavor.