Baseball for the Thinking Fan

Login | Register | Feedback

You are here > Home > Primate Studies > Discussion
Primate Studies
— Where BTF's Members Investigate the Grand Old Game

Friday, November 09, 2001

Quibbling with the Quibblers

To bunt, or not to bunt: that is the question.

Some Post-World Series Post-Mortems

The World Series is a lightning   rod for what I?ve taken to calling the ?armchair army,? that vast subgroup of   baseball fans who cover the spectrum from the sabermetric know-it-all to the   ?dese and doze? insider apologist.

Because this is the one time   during the year when all eyes are at least semi-focused on baseball, the level   of verbal warfare and ?high dudgeon? is at its peak, often exceeding the routine   excesses of the media.

In other words, we all become engrossed in the not-so-gentle art of verbal-tribal   warfare that accompanies the on-field action. With matters boiled down to a   simple choice of two teams, we can fix our attention on the discrete events   unfolding in an impossibly small number of games, and attach great? (read: overweening)   significance to each of these details.

The past has shown us that
  this is the accepted approach, for history is most easily constructed by exaggerating
  the meaning of one event and extending its effect over an entire era. Baseball
  is by no means an exception to this precept.

The World Series that appears
  to have the most similarity to this one?the 1960 epic between the Yankees and
  Pirates, in which Pittsburgh survived three drubbings to win Game Seven in their
  last at-bat?produced one of those so-called ?epochal moments,? when Casey Stengel
  was fired after winning ten pennants in twelve years.

Those are the kind of signposts that media types dote upon?and   it?s not completely a sure thing that current Yankee manager Joe Torre, having   committed the unpardonable sin of losing a World Series, will escape the fate   that befell Stengel.

Back to the key point, however,
  which was second-guessing as the


?national pastime.? This is the
  time of year when managerial performance is subjected to intense (read: nose
  hair-level) scrutiny, and many find the essentially conservative nature of post-season
  play to be at odds with the way the game is often played during the regular

Arizona manager Bob Brenly
  has been second-guessed more thoroughly and from all ideological vantage points
  than any manager in recent memory. From the ?insider? world, he was criticized
  for giving Curt Schilling three starts (by an ?unnamed fellow general manager,?
  whose identity was protected by that paragon of journalistic discretion, Murray
  Chass (rhymes with?well, you


what it rhymes with?) of the New
  York Times

By forcing Schilling into
  three games, so the theory went, Brenly was guaranteeing that he?d have to go
  into his bullpen in those games, where dangerous frailty resided.

The fact that complete games
  are exceptionally rare, even in the post-season, appears to be lost on this

Keep in mind that these statements
  were quoted by Chass only after the Yankees had staged two consecutive two-out,
  ninth-inning rallies, something that hadn?t occurred since 1929 and whose incidence
  can still be counted on the fingers of one hand.

This probably counts as the
  ultimate insider shaggy-dog second-guess, made by someone whose team wasn?t
  even in the World Series. Perhaps Chass knew it was puerile, and protected his
  source from public embarrassment by concealing his identity.

(On the other hand, Chass
  is also the guy who thinks that the owner?s plan to eliminate two teams is ?intriguing.?)

Over here at the home of the ?disloyal opposition,? Brenly was hammered hard   in Alan Shank?s Tidbits blog for the opposite reason.? In general, sabermetricians   despise one-run strategies, and this tendency is quite pronounced in Al?s interpretation   of the events in Game Seven.

Al had a veritable laundry
  list of things that Brenly did wrong, but the key ones appear to be: 1) he left
  Schilling in too long and didn?t bat for him in the bottom of the seventh in
  Game 7; 2) he initially played for a tie in the bottom of the ninth and didn?t
  use his bench well.

I?m not here to apologize
  for Brenly, but I think we might want to look at some other contexts here just
  to see what shakes out of the tree.

First, let?s establish the ambient conditions of post-season   play and see how these compare to those of the regular season.

Post-season games have a
  strong tendency to be tighter and significantly lower-scoring than games in
  the regular season. From 1901-2001, the average number of runs scored in the
  World Series is 7.4 (both teams); in the regular season, that average is 8.7.

That?s a 15% reduction in
  run scoring. The 2001 World Series, despite two lopsided wins by Arizona, was
  no different: 51 runs were scored in seven games, or just under 7.3 runs per

From 1901-2001, there has evolved an almost perfect bell-curve structure in   ?run scoring differential,? or the number of runs by which a game is decided.   49% of all games played since 1901 have been? ?close games? (decided by two   runs or less); 25.5% have been decided by 3-4 runs; and 25.5% have been ?blowouts?   (decided by five runs or more).

In the World Series, there have been more close games and fewer blowouts, as   the accompanying chart? (organized by decade) reveals. This goes hand-in-hand   with lower run scoring, resulting from the concentration of better pitching   on post-season teams (and the tendency to use the best pitchers on these teams   more frequently in a short series).

It is this set of ambient
  conditions, which look more like deadball-era or 1960s-style baseball than what
  we have been used to in the past nine years, that dominate how decisions get

(Oddly, you?ll note that
  in the 60s there were a very high number of ?blowout? games in the World Series,
  which is counterintuitive in terms of what we?d expect to see. This is a classical
  example of ?small sample size aberration? at work.)

So, as a manager, you are
  already working in an environment of reduced run scoring. And, in Brenly?s case,
  you are in a Game 7 (do or die, in other words) in a 1-1 tie.

Al makes a couple of points about the Schilling decision, each of which tends   to cancel the other out. He notes that Schilling had given ground in the seventh,   and allowed the tying run to score. But he then notes that Soriano?s home run   in the top of the eighth was on a ?pitcher?s pitch.?? I take this to mean that   Schilling was still making good pitches; but, as Jim Bouton says, ?sometimes   hitters hit homers on good pitches.?

(It?s funny how people see
  the same pitch in such different ways, of course. Rob Neyer, in his own shellacking
  of Brenly, suggests that Schilling threw a bad pitch to Soriano. The replays
  show, however, that Soriano hit a pitch below his knees. Like many, Neyer has
  his own elaborate network of second-guessing, all of which is flawed by the
  fact that a theoretical argument is being floated


the actual game
  context is known.)

So at best we have a mixed message here. I think it?s highly likely that if   Brenly had batted for Schilling in the seventh, and Arizona hadn?t scored, and   the new pitcher had allowed the go-ahead run(s) in the eighth, the ?insider?   contingent of second-guessers would have come out of the woodwork like lemmings   rushing headlong toward the seacoast.

So, essentially, however   you choose to sort out which move you?d make probably indicates more about your   own predispositions with respect to this matter than anything else. There is   no one ?correct? choice.

So then we get to the bottom   of the ninth, with the D-backs trailing 2-1, facing Rivera. (One thing I haven?t   seen commented on, by the way, is whether Rivera has any tendency to give up   runs after so many consecutive appearances where the outing is longer than an   inning. There?s game data available to us that might address that question,   and I leave it to some enterprising type out there to compile it.)

Al?s next issue with Brenly   is based on what to do after Mark Grace opens the inning by reaching base (with   a single). Do you go for the big inning against Rivera, knowing you have only   two outs to give, or do you spend an out in order to try for a tie? Al opts   for the Earl Weaver option, while 98% of all managers in such a situation take   the Gene Mauch option.

All I can say is that if   I?m facing Rivera in this situation, I?m going to play to at least get the tie.  

A major side issue in Al?s discussion concerns the non-use   of Erubiel Durazo. Later on in his discussion, Al references the reason why   Durazo (the darling of sabermetric types in large part because he?s been blocked   from a starting job by an older and arguably inferior player) wasn?t used: he   represented a greater double play threat.

But he does so only after
  beating up Brenly three or four times for not using him in situations where
  the double play would be catastrophic to the Diamondbacks? fortunes.

In the context of playing for one run (man on first, no   one out), Erubiel Durazo is not the man you?re going to use to lay down a bunt.

However, if you bat Durazo
  and he hits a grounder, you?ve just put yourself within one out of extinction.

While the one-run strategy
  appears to be excessively conservative to sabermetric types, it?s the play to
  use, because if the defense screws things up, you?re in a much better position
  to get the tying


winning runs into scoring position with less than
  two out.

That?s exactly what happened,
  when Rivera threw wildly to second. Al referred to run expectancy tables that
  show only a modest increase in run scoring likelihood as a result of a successful
  sacrifice, but he doesn?t discuss the scenario where both men are safe.

The chances of scoring the
  tying run are nearly tripled when such occurs.

With this situation (down one run, ninth inning, none out,   runners on first and second), Al apparently agrees that the one-run strategy   becomes more viable. If we assume a successful sacrifice at this point, we can   anticipate that Torre will walk Tony Womack to load the bases, bringing Craig   Counsell to the plate with the bases loaded and one out.

Then, of course, Brenly got unlucky when his most accomplished   bunter, Jay Bell, bunted into a force play.

It?s at this point?one out,
  runners on first and second?where Al calls for Durazo to pinch-hit for Tony
  Womack. But again, the spectre of the double play is a valid reason for not
  making such a move.

A ground ball by Durazo in this situation, in all likelihood,   means the end of the game.

Womack is not nearly as good
  a hitter, as Al rightly points out. But part of what must be assessed in this
  situation is

what type of an out he might make

(since the chance of either
  doing so is higher than 50%), and

whether he has a greater or lesser risk
  of making two outs


Womack is far less likely
  to make two outs in this situation than Durazo.

(Interestingly, the stats tell us that Durazo doesn?t hit into many double   plays, in part because he hits fewer ground balls than the average major league   hitter.? But the stats can?t tell us what will happen in this particular at-bat,   so risk aversion is a legitimate perspective in this instance. There?s also   this fact: when Rivera is not striking out batters, he?s allowing more than   twice as many ground balls than fly balls.)

Weighing all of this, I?d have to conclude that Brenly made the right choice.?   He then got lucky when Womack pulled a pitch down the right field line for a   double, tying the game and putting the winning run on third with less than two   out.

Al then wants Durazo to bat
  for Craig Counsell. I can see that, but I can also see Joe Torre simply walking
  him, since his run means nothing. If the opposition manager can bypass your
  best pinch-hitter, why wouldn?t he? So why should a manager risk allowing his
  best pinch-hitter to be bypassed in such a manner?

In addition, Brenly needs to keep tabs on his personnel?specifically his supply   of middle infielders.? Bell, used as a pinch-bunter, would still be available   to play second if you batted for Counsell, but you?re not in a spot where you   can bat for Womack and Counsell.

In short, most of this criticism
  is nothing more than a somewhat more mathematically enlightened version of armchair
  quibbling. I?m not convinced that the alternatives presented are definitively
  superior to what Brenly actually did, and the subtext (that Brenly is an idiot)
  is more of that relentless ?insider bashing? so popular in the little world
  of the sabermetrically aligned.

Sabermetrics has much to
  offer baseball analysis, but most of its insights occur at a ?macro? level?general
  principles applying to a part or all of a season. The field has made very little
  real inroads into the ?micro? level?the world of the individual game, or the
  sub-world of the ninth inning in Game 7 of the World Series.

Run expectancy tables do
  some of that for us, and they are a valuable addition to our knowledge base.
  Al applies them well?albeit somewhat selectively. But they do not constitute
  an inviolate, cut-and-dried ?decision table,? either.

The drift of sabermetric
  discourse from asking questions about important and little-documented aspects
  of the game toward a ?we?re smarter than they are? mentality has its roots in
  that portion of Bill James? work that can now clearly be seen as his least valuable.
  Unfortunately, it?s that portion of James? work that has been most imitated
  by those who have followed him.

Bob Brenly did something
  semi-radical in this World Series. He went back to a type of starting pitcher
  usage that had been abandoned as a result of a kind of blind ?conventional wisdom?
  that had grown over the game like a persistent weed over the past twenty years.
  He ought to get some credit for taking such a step, as part of any assessment
  of his performance as manager?whether it is tilted from an ?insider? or a ?sabermetric?

Quibbling, however, remains a more self-satisfying practice?for all elements
  of the ?armchair army.? As a result, a truly balanced assessment of Brenly?s
  performance continues to elude us.


Don Malcolm Posted: November 09, 2001 at 05:00 AM | 12 comment(s) Login to Bookmark
  Related News:

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

   1. Voros McCracken Posted: November 10, 2001 at 12:13 AM (#604202)
"Womack is not nearly as good a hitter, as Al rightly points out. But part of what must be assessed in this situation is what type of an out he might make (since the chance of either doing so is higher than 50%), and whether he has a greater or lesser risk of making two outs."

Womack is far less likely to make two outs in this situation than Durazo."

This year Womack grounded into 4 Double Plays in 518 PAs. In his career, Erubiel Durazo has grounded into 5 Double Plays in 625 Plate Appearances. As a leadoff hitter in the national league, Womack also has a far greater percentage of his PAs without runners on than Durazo does, so Womack probably has had relatively fewer Double Play opportunities per at bat than Durazo (I checked and he has batted much less with runners on). Womack is much more of a ground ball hitter while Durazo hits lots of flyballs.

Womack is faster than Durazo, but I don't see evidence that he'd be "much" less likely to ground into a Double Play than Durazo.

I'm not going to discuss whether Durazo should have hit for Womack (there are points in Womack's favor), I simply wanted to address that point as I think you've made an unsafe assumption. Durazo does not appear to be a particularly inviting double play target. Luis Gonzalez is much more of one. Perhaps Duraco should have batted for him. :)
   2. Alan Shank Posted: November 10, 2001 at 12:13 AM (#604208)
A couple of things, Don.
1. In recognition of the lower scoring we generally see, and certainly this year, in the postseason, I used the base/out table from the "pre-silly-ball" era, rather than the one I normally use based on '95-'98 or some such period.

2. It's true that for a full analysis of sacrifice you need to consider all the possible outcomes. My main point there was with the "successful outcome." Assuming the sacrifice "works," you increase the chance of scoring one, and tying the game, but decrease the chance of scoring two and winning. I felt that Arizona would be at a disadvantage if the game went extra innings. I went on to mention that if the sacrifice fails, you throw away an out for nothing. This is probably more likely than what actually happened, the throwing error. I would like to see a study of play-by-play data for a whole season and find out the frequency distribution of:
sac fails, lead runner out
error or FC, all hands safe
I would be surprised if 3 is as common as 2.

"Al makes a couple of points about the Schilling decision, each of which tends to cancel the other out. He notes that Schilling had given ground in the seventh, and allowed the tying run to score. But he then notes that Soriano?s home run in the top of the eighth was on a ?pitcher?s pitch.? I take this to mean that Schilling was still making good pitches; but, as Jim Bouton says, ?sometimes hitters hit homers on good pitches.?"
I don't think those point cancel each other out at all, Don. In the 7th, Schilling gave up three line-drive singles and got out of the inning on a nice catch by Finley on a deep fly by Shane Spencer. I mentioned that the pitch Soriano hit for a homer in the 8th was not a bad pitch so as not to imply that it was a bad pitch. Had Brenly pinch hit for Schilling in the bottom of the 7th, obviously we don't know what would have happened. When you're in game 7 of the World Series, you're tied, you have one inning before Mariano Rivera is probably going to enter the game and your pitcher, who, based on that last inning, appears to be losing or have lost his effectiveness, is leading off, I think it behooves you to pinch hit for him. We all know the importance of getting the leadoff batter on in an innging.

4. Womack vs. Durazo. As Voros has pointed out, Durazo did not hit into more DP than Womack. Womack is faster, to be sure, but also hits a higher percentage of grounders than Durazo does. The biggest difference, of course, is that Womacks "out percentage" is much higher than Durazo's.

5. I never criticized Brenly for pitching Schilling three times; in fact, I based my probability spreadsheet on the assumption that he would. I did think he should have pulled him earlier in game one, and Johnson in game six.

Alan Shank
Al's Baseball Tidbits
   3. Robert Dudek Posted: November 10, 2001 at 12:13 AM (#604210)

Is it second guessing if you make your statements before the result are in?

One of the things that Scruff and I did on our website ( is that we provided in-game commentary for all playoff games. The time stamps are not faked, so it is possible for anyone to check the statements we made before the fact.

In the future, I am certain that more and more journalists/analysts will provide "real-time" analysis and so those that do not won't have the convenience of saying "I told you so" after the fact. The proper response will be - why didn't you say anything before it happened?

A few points. Having 2 pitchers like Johnson and Schilling who are so much better than the other pitchers on the staff, how can you NOT give them 5 starts in a 7 game series? In 1991, Minnesota starters pitched games 4,5,6 and 7 all on 3-days' rest. I can see no justification in doing anything different. Maybe Johnson should have started games 1,4 and 7, but this is a comparably minor point. Leaving Schilling in longer than necessary was an obvious mistake (same could be said for Johnson in game 6 - even though that turned out well).

The question of when to take Schilling out in game 7 was a difficult decision and I think on balance it would have been better to pinch-hit for him. But there are pychological factors at work: if you don't score you go to the pen earlier, which, given its shakiness, would probably make it more difficult to win an extra inning game (fewer pitching options). The pitch Schilling threw to Soriano was a very good pitch - exactly the kind you'd want to throw to a free-swinger (out of the zone with downward movement). On the other hand, Schilling threw almost the exact same pitch earlier in the AB and Soriano fouled it off instead of striking out.

The key mistakes that Brenly should have been lambasted for were his handling of Kim in games 4 and 5 and of Schilling in game 1.

   4. Kurt Posted: November 12, 2001 at 12:13 AM (#604220)
Just a couple of points:

1. Don, glad to see you back.

2. I've seen plenty of people, here and elsewhere, give credit to Brenly for starting Schilling 3 times. I don't think it deserves an inordinate amount of credit because it's a fairly obvious move, but making the obvious move is still better than not making it.

3. I don't have a big problem with playing for the tie in that situation. It's presumably Rivera's last inning, and the Yankees' other relievers are a lot worse than Rivera, so getting the game to the 10th would be very appealing to the D'Backs.

4. Any analysis of the sac bunt needs to consider *all* the options - successful sacrifice, unsuccessful sacrifice, everybody's safe, and everybody's out (via DP). The bunt DP is not so rare as to be safely discarded from the analysis (it almost happened on the second bunt) and a DP virtually ends the game.

I agree with Don that the first bunt was not so horrible that Brenly should be ripped for it. I still find other decisions (not getting Durazo in, leaving Johnson in in Game 6) pretty hard to defend.
   5. scruff Posted: November 12, 2001 at 12:13 AM (#604222)
1. " Al then wants Durazo to bat for Craig Counsell. I can see that, but I can also see Joe Torre simply walking him, since his run means nothing. If the opposition manager can bypass your best pinch-hitter, why wouldn?t he? So why should a manager risk allowing his best pinch-hitter to be bypassed in such a manner? "

He's not going to walk him to get to Gonzalez, and if he does, with the bases loaded, and no margin for error (meaning Rivera MUST come with a pitch in the zone, or risk walking in the winning run), I'd take it. That's better than Counsell making a non-runscoring out and leaving Gonzalez up there with 1B open. Brenly got lucky that Counsell was HBP, when the replay clearly shows he was not hit.

2. I have absolutely no problem with the first bunt (after Grace's single) makes all the sense in the world to play for 1 run in that situation. I disagree with the 1st and 2nd bunt. You've already got the tying run in scoring position. I strongly believe you take 3 shots at getting that run home. The force at 3rd is nearly as easy a play as the force at 1st in that situation, which is exactly what happened.

3. You have to send Erubiel up for Womack w/runners on 1st and 2nd and one out. I don't see any defense for anything otherwise. Who is more likely to single and tie the game? Double and possibly win the game? HR and win the game? Who is less likely to hit into a DP? Who's a better defensive SS if you only tie the game, Jay Bell or Tony Womack?

Only if you don't use Erubiel there, do the arguments come about for not sending him up for Counsell. Obviously no one is going to PH for Womack and Counsell with the same guy. I would prefer to see Durazo-Counsell or Womack-Durazo, but Womack-Counsell? I don't get it.

As for the roster composition, do you mean to say there was no middle infielder in the Arizona organization that runs as fast as Midre Cummings? Not having an extra MI is Brenly's fault as well. You know you have three excellent hitters sitting (or rotting) on your bench. You have a couple of light hitting middle infielders at the top of your lineup. Why not keep a second middle infielder instead of a 6th OF as your pinch-runner? What the hell was he thinking?

As for Schilling, I thought it was a mistake to lift him in Game 4, and the right call not to PH for him in Game 7. Soriano hit a great pitch, what are you going to do? I don't blame Brenly at all for sticking w/Schilling in that situation. I absolutely agree that you cannot knock him for Game 4 and ride him for Game 7. He botched his staff enough in this Series that we don't have to pile on. He gave us plenty to criticize legitimately.

I do disagree with bringing in Kim for 2 innings in Game 4 with 9 days off and a TWO run lead. Send Schilling back out there and if he gets in trouble go to Kim. Or if Schilling is gassed, send Morgan out for the bottom of the order and if two get on go to Kim. Relief pitchers are generally more erratic after long periods of inactivity, so minimize Kim's outing if possible, especially since you have another game tomorrow without one of your two horses (Batista is a very good pitcher, but he doesn't complete too many games).

Predictably, by the time Tino hits his dinger, Kim is around pitch 40, which could have been expected in a 2-inning performance from a sometimes wild strikeout artist. And it's inexcusable to leave him out there for the 10th. And even more inexcusable to then bring him back the next day after throwing 62 pitches? What the hell could Brenly have possibly been smoking?

In the New Historical Abstract, James actually writes up a study comparing reliever usage patterns over time, and he gets at the best scenario. His conclusions are intuitively very sound, in that the 2nd and 3rd run prevented by a closer are of relatively minimal value, the first run is most important. Even bad pitchers don't blow 2 and 3 run leads very often.

He sites optimal closer usage as:

-2 innings w/a 1-run lead
-2 innings in a tie game
-1 inning when the game is close and closer hasn't pitched in a few days.

The third option fits perfect here, and I wrote a similar comment during the game, before I had even read the James Article (which is found in the 1950's or 60's section of the book). This is very similar to Joe Torre's postseason usage of Rivera, although James endorses this for regular season as well. He says it would result in about 69 G, 113 IP seasons for closers, which is a reasonable workload in my opinion (and his).

I also disagreed w/leaving Randy out for the 9th of Game 7, he looked very hittable in the Knoblauch AB (last out of the 8th) and I thought he was gassed. He was topping out around 92 MPH on his fastball. But that move worked, more power to Brenly there.

These were my views during the game, there is no revisionist history at work here or anything. The bottom of the 9th for Game 7 isn't entirely posted on Mostly Baseball, because I am a Yankee fan and I was in the middle of a complete nervous breakdown and things were happening too quickly for me to type them all in.

But they were my thoughts at the time. Really, that whole premise of revisionist history for our bottom of the 9th, Game 7 strategy criticism doesn't make sense, because the moves worked. If we didn't truly believe Brenly goofed, we wouldn't mention it. Obviously we think the are SO BAD that we still criticize them, even though they worked. We are talking about a guy that tried to suicide squeeze home the winning run in the bottom of the 9th of Game 5 of the Division Series Series. With one out. My exact words were:

"A SUICIDE SQUEEZE??? WHAT COULD HAVE POSSIBLY BEEN GOING THROUGH BOB BRENLY'S SKULL??? A wild pitch could have scored the winning run, you had two chances at a hit. That is just an atrocious case of overmanaging.

Now WP doesn't score him and you need a hit from Womack, against a lefty. Why isn't Erubiel Durazo batting here? Awful."

We were "wrong" in that situation too. As I've said many times over the last month, even a blind squirrel finds a nut every once in awhile. Or in Brenly's case a bushel full of nuts.
   6. Darren Posted: November 12, 2001 at 12:13 AM (#604223)
You arguments are not very convincing. You defend keeping Durazo on the bench as "risk aversion." But you don't provide any facts on the risks involved. Your position actually flies in the face of the facts.

You don't pinch hit Durazo for Counsell, because they might walk Durazo? Well that means that you've got Gonzalez and Williams hitting instead of Counsell and Gonzalez. And you're guaranteed of Gonzalez batting with only 1 out, keeping the sac fly in order.

And are you arguing that the bunt is a good play because of the chance of an error occuring? How often does that happen? When you consider how often it happens, does it make the bunt a better percentage move? You can't just say "and that's exactly what did happen" to prove your point.

   7. Alan Shank Posted: November 12, 2001 at 12:13 AM (#604226)
"3. I don't have a big problem with playing for the tie in that situation. It's presumably Rivera's last inning, and the Yankees' other relievers are a lot worse than Rivera, so getting the game to the 10th would be very appealing to the D'Backs."

It wasn't going to be the Yankees'other relievers, but Mike Mussina that would have been pitching after the 9th, which is one of the big reasons I felt Brenly should have managed to win in the 9th.
Alan Shank

   8. Don Malcolm Posted: November 15, 2001 at 12:13 AM (#604231)
Some thoughts (yes, MUCH longer than I anticipated) in response to the comments thus far?

First, let me repeat that what?s most interesting to me here is the gap between ?macro? and ?micro? analysis in baseball, which is often (but not always) an opposition between ?sabermetrics? and ?insider? perspectives.

That?s where we need to focus our efforts, IMO?in looking for ways to synthesize (or, failing that, reconcile) these points of view.

What has happened thus far in the comments, as I read them, is that we have simply shifted the quibbling to different ground. We are never going to eliminate second-guessing, but we can create a framework for it that might result in some more open ears from the vast majority of baseball fans who have yet to absorb the lessons of the ?macro? perspective.

As I noted, commentators from a ?sabermetric? perspective were (and remain) adamant that Durazo appear as a pinch-hitter for SOMEONE in the ninth inning. After thinking about this some more, I?ve come to the conclusion that the only truly strong argument for this is located at the point after Rivera?s throwing error, when Bell was sent up to sacrifice.

I think that this is a legitimate point of contention on the part of the ?sabermetric? contingent, but when I say that, I don?t mean that one has free rein to call Brenly?s decision to bunt a mistake.

This is one of those ?Rubicon moments? in the game, and the run expectancy tables are only one point of guidance.

Even though Al has indicated that he was using pre-1995 tables, we need to remember that those figures are aggregate data, and do not take into account the following:

?quality of pitcher on the mound
?general quality of opposition
?the specific inning involved

I?ve never seen run expectancy tables generated by specific inning, and I?d be interested in seeing how they differ from one another. That would be a breakout that someone like Tom Ruane could supply us, I?d guess (and, yes, if you?re out there reading this, Tom, that?s definitely a hint?).

The bottom line in my mind is that IF you are going to play to win in this situation (first and second occupied with no outs), the most optimal time to do so is at the point when you?re batting for the pitcher. WHETHER OR NOT you should do so, however, is still an open question.

The main reason for bunting a second time is to get the tying and winning runs into scoring position with less than two out. Also, by doing so, you eliminate (or, at least, greatly minimize) the chances of a GIDP. Let?s do an alternate universe scenario, and take a look at what happens if Bell?s bunt is good and the sacrifice works:

?Womack bats with one out and runners at second and third.
To get the force play at the plate, Torre has to walk him.

?Brenly then can bring Durazo up in the spot where he wants
his best bench hitter to come up?down a run, needing a fly ball
to tie, with a base hit a likely game-winner.

The failure of Bell?s bunt, of course, leads to the next crossroads. Voros has supplied the exact numbers (to which I alluded, rather than supply initially, so that some enterprising person would do some actual research on the matter?) on the issue of Womack/Durazo and double plays, which we'll revisit in just a minute.

This is the most crucial moment in the Series, of course, and it certainly appears that Brenly went with his gut. My sense is that he never thought about batting for Womack, who?d already had a clutch hit in the NLDS against the Cardinals. Even though sabermetricians dismiss such things as random events (and yes, they?re correct), managers tend to remember them.

(Has anyone seen any article where a member of the media actually asked Brenly about his decision-making in the inning?in general, or specifically about his non-use of Durazo? That could make for interesting reading.)

The thing about sabermetrics when it wears its ?micro? analysis hat is that it dotes upon this type of situation?critiquing the manager for going against the better statistical odds, and sticking with that critique even when he beats those odds.

That nine percent difference in out percentage between Durazo and Womack is not so large that one can hang his hat on it. To his credit, Voros avoids such a position, though he doesn't provide any actual reasons for doing so.

But his point about double play situations is shakier than it seems at first glance, because we don?t have the exact number of times either Womack or Durazo batted in ?double play situations.? What I DO know is that when Durazo hits a ground ball, he hits into more double plays than Womack. And I also know that, as noted in the article, Rivera has a strong tendency to produce grounders.

The figures: in 2001, Womack hit into a double play once in every 53 ground ball outs. In his career, Durazo has hit into a double play once in every 34 ground ball outs.

So, based on that, I suggest that Womack still has a significantly better chance of avoiding a GIDP if he hits a grounder against Rivera.

We could probably compute the odds of how likely each man was to hit a ground ball more definitively, and if someone wants to crunch those numbers in order to refute my contention, that?s OK too.

Remember that I said that "risk aversion" has to be part of this decision, and in a way that is far more compelling and prominent than in a game during the regular season. You're two outs away from losing the World Series; one of the factors that the manager MUST consider is avoiding the double play.

The other reason not to bat Durazo in that spot is that his tendency to hit fly balls doesn?t help you get a run in after Bell?s bunt fails (runners are only at first and second at this point).

Now let's shift ahead to the point after Womack's game-tying double, and address some of the comments clustered around the Durazo vs. Counsell issue...

Those who question the notion that Torre would have intentionally walked Durazo had he batted for Counsell seem to forget the situation that the Yankees faced at that point. The game was tied, with only one out. The standard play here is to load the bases in order to create a force play at the plate. In this instance, it really doesn?t matter who the next batter is.

By focusing solely on the idea that Brenly has to play to win, Al (and others here) appear to simply dismiss the idea that he should even consider what personnel he?ll have available in later (extra) innings. The assumption that the Yankees would have an advantage if we go to extra innings is just that?an assumption. Mike Mussina is an excellent pitcher, but he was pretty mortal in the World Series.

(And actually, Torre might well have waited until he got through the tenth to bring in Mussina, since the pitcher?s slot was due up fourth in that inning. If you?re going to use Mike in the game, you might want to get as many innings out of him as possible.)

However, if you stop to think about it, having Durazo available in extra innings to bat against either Mussina or Ramiro Mendoza isn?t exactly a bad idea.

(Quick thought: someone out there should study Brenly?s use of his bench in 2001?for the whole season, not just the World Series. It might prove interesting.)

What?s surprising to me is that, to my knowledge, no one has suggested that Torre may have pushed Rivera too hard by giving him so many multiple-inning stints. James? study of relief pitcher usage in his new book is quite valuable, but like much of his post-Abstract work, it seeks a much higher level of generalization and thus doesn?t address more individualized usage patterns that require examination before his theory can be deemed sound. One such pattern is relevant to our discussion here: whenever Torre has given Rivera three or more outings in six or fewer days with an inning or more in each, the result has been similar to what we saw in Game Seven.

Over the past four years, Torre has done this seventeen times. And in that third or fourth outing (not counting post-season appearances), Rivera?s ERA is 6.35 (13 ER in 181/3 IP).

There?s another interesting stat that comes out of this. Rivera has appeared in more than one inning 66 times over the past four years, as opposed to 191 appearances of an inning or less. While his ERA is lower in those multiple inning games (2.03 to 2.40 for <= 1 IP), his save percentage in multiple inning games is a good bit lower (43 of 58, or 74%, as opposed to 124 of 130, or 95%, in single-inning appearances).

Now does this mean that I want to take Joe Torre to task for using Rivera the way he did? No. I?m simply commenting on the one-sidedness of this discussion.

Those who are so quick to judge Brenly over his staying with Schilling have evidenced no problem with Torre?s actions. In reality, however, both men were simply going with the pitcher they considered to be their best man.

And by doing so, they were attempting to insulate themselves from ?insider? second-guessing?especially second-guessing from within their own organization.

Think about it. If Torre had gone to Ramiro Mendoza in the eighth inning and Mendoza had surrendered the lead, the backlash would have been instant and brutal, centered around the precept that ?you have to have your best man out there at the big moment.?

As a matter of fact, it?s not too much of a stretch to surmise that had Torre lost the Series with Mendoza pitching in the eighth, he would have been fired a la Stengel in 1960.

(By the way, the 1960 World Series is interesting from an analogous reliever usage standpoint. You are invited to examine Danny Murtaugh's use of Roy Face, while paying attention to the "innings/outing/days of rest" continuum.)

A similar type of ?insider? backlash would have greeted Brenly had he removed Schilling for a pinch-hitter in the seventh, and his reliever had surrendered the go-ahead run to the Yankees in the next inning.

And I submit to you, with all due respect, that to the extent that a manager (Brenly or Torre or anyone...) is concerned with second-guessing, he's far more worried about ?insider? backlash than he is about ?sabermetric? backlash.

I?m NOT saying that this is best way to make decisions. I?m merely noting that other forces operate on those decisions, and if we are going to be fair about this, we need to take those forces into account. Frankly, I?ve seen too little of that in this discussion, which is what prompted me to write the article originally.

Bill James? idea that a top reliever can pitch two innings in a game on consecutive days, or other similar workload patterns, is entirely an abstract construction at this point in time. It will take some serious, detailed historical research into the specific usage patterns of relievers in the pre-Bruce Sutter era to draw any conclusions about that. As a general theory, I think it makes a good bit of sense?but there are always going to be exceptions, and a detailed study of past usage similar to what he is proposing would be a welcome addition to our knowledge base.

Much of what happens in the World Series is indeed the ?randomness? that has become the most recent buzzword in sabermetrics. We make too much of these small sample size events at the peril of our own credibility in a world where most of the people haven?t digested those precepts.

What may have guided the Diamondbacks to victory is another ?random? element in the probabilities: the inherent home field advantage that the best-of-seven format creates. The teams who have the home field advantage have won 55 of 97 World Series (56.7%). That exceeds the lifetime home team advantage (they?ve won 54% of the time since 1901).

The Yankees? record in the World Series when they?ve had home field advantage is a spiffy 16-5 (which accounts for most of the historical advantage for home teams). However, they?ve also contributed to the success of teams starting the Series on the road, going 10-7 in those Series. When you take the Yankees out of the World Series equation, the number of Fall Classics won by the team with home field advantage is 54.2%?virtually indistinguishable from the historical home team winning percentage.

I also think that the nature of the Diamonbacks? loss in Game Five (d?j? vu all over again?) has colored some of the commentary, at least subliminally. When a team loses one game in this fashion (as in Game Four), the natural reaction is ?bad luck.? When it happens two times in a row, the natural reaction shifts to ?bad manager.? From that point forward, the magnification power of that microscope Brenly was under enlarged tenfold.

Due to the length of all this, I?m not going to address any of the other ?second guesses? that were tacked on by some of the respondents, except to note that I haven?t seen any documentation for the claim that closers are ?more erratic? after a long layoff. If anyone has it, please post it.

(A final parenthetical note: if you want to see some REAL second-guessing, check out this David Letterman-style ?top ten? list circulating in St. Louis as a result of the Cardinals? loss in Game Five?courtesy of Nic Antoine, aka SABR?s ?dark knight?:

10. Well, you don't want to run on Mike Piazza.
9. It's the National League?don't the pitchers HAVE to hit?
8. C'mon, Craig Counsell's never had a big hit in his life, there's no reason to throw my best guy against him.
7. I think Miguel Cairo is an ideal #3 hitter.
6. They don't have vegetarian restaurants in Atlanta. [Tony LaRussa, the target of this list, is purportedly a vegetarian, though some have claimed that he ?eats meat on the road.?]
5. [Eli] Marrero is great at stealing signs, so I need him on the bench.
4. [Matt] Morris had only thrown 110 pitches...might as well leave him in, since he hasn't had arm problems in a while.
3. I needed someone who could hit the ball out of the park, and I looked down and there was Kerry Robinson.
2. You don't want to nibble with Tony Womack at the plate.
1. [Craig] Paquette has pictures of me torturing a cat.

I?m sure that the ?glib contingent? among the second-guessers out there can ?do? Bob Brenly in this manner if you so desire?)
   9. Kurt Posted: November 15, 2001 at 12:13 AM (#604237)
More quibbling...

"But his point about double play situations is shakier than it seems at first glance, because we don?t have the exact number of times either Womack or Durazo batted in ?double play situations.? What I DO know is that when Durazo hits a ground ball, he hits into more double plays than Womack. And I also know that, as noted in the article, Rivera has a strong tendency to produce grounders.

The figures: in 2001, Womack hit into a double play once in every 53 ground ball outs. In his career, Durazo has hit into a double play once in every 34 ground ball outs.

So, based on that, I suggest that Womack still has a significantly better chance of avoiding a GIDP if he hits a grounder against Rivera."

I don't think that analysis is sound, for two reasons:

1. It doesn't tell us how often Durazo and Womack hit into ground ball outs. Obviously if each hits a ground ball Womack is more likely to beat it out, but the critical question to me is "how likely is a DP for each player?" which has more to do with whehter a ground ball is hit than how many ground ball outs result in DP's.

2. As you noted, it doesn't tell us how often they come to the plate with the DP in order. If Womack generally bats leadoff, in front of the pitcher, and Durazo bats in the middle of the order, ot would stand to reason that more of Durazo's ground ball outs result in DP's. This year Womack had 504 plate appearances; 178 with runners on. 35% of his PA's came with runners on. Durazo had 303 PA's, 110 with runners on, for 54%. Of course more of his ground ball outs will result in DP's.

That's as far as I can go, though, I don't have the exact DP opportunity figures.

"The other reason not to bat Durazo in that spot is that his tendency to hit fly balls doesn?t help you get a run in after Bell?s bunt fails (runners are only at first and second at this point)."

Sure, but it helps avoid the DP. Besides, with runners on first and second there are virtually no "productive outs" (other than the sac bunt), so the question is who is more likely to not make an out.

"What?s surprising to me is that, to my knowledge, no one has suggested that Torre may have pushed Rivera too hard by giving him so many multiple-inning stints."

I had thought about this, but I think this falls under the sabermetric aversion to second-guessing. Given the level of success Torre has had over the last 6 years putting Rivera in for 2 inning stints, it seems like a classic second guess to question it now. Although analysis would of course be useful.

"And I submit to you, with all due respect, that to the extent that a manager (Brenly or Torre or anyone...) is concerned with
second-guessing, he's far more worried about ?insider? backlash than he is about ?sabermetric? backlash. "

No question. Not that that'll stop us or anything.
   10. Don Malcolm Posted: November 15, 2001 at 12:13 AM (#604240)
OK, Kurt, you caught me. We have a better (but still not quite definitive) source for this data that can get us further along. And yes, I didn't think it through properly. My bad.

The most detailed situational data for hitters I've seen (which can be found at breaks out situations by runners on base (1st, 1st/2nd, 1st/3rd, 2nd only, 3rd only, bases loaded), but it doesn't give us the number of outs.

So we don't know how many times these guys batted with men at the various bases with less than two out (aka the DP situation).

However, we can figure out that 44% of Womack's at-bats result in ground ball outs, while Durazo's percentage is 32%.

If we prorate GB/FB in the exact proportions to the overall totals (which is clearly not going to be statistically accurate), we can project the likelihood of a DP based on the number of DPs in the "target DP situations" (1st, 1st/2nd, 1st/3rd, bases loaded).

When we do that, we find that Womack still comes out as less likely to hit into a DP, but the difference is much smaller (down to between one and two at-bats difference per DP).

In that case, I'd have to say that this argument is not particularly strong, and that there is certainly a viable POV supporting batting for Womack with Durazo.

Again, however, if the 240--lb. guy happens to hit into the DP, the "insiders" will bury you.

One piece of research that should be done and made available (maybe as part of yet another upgrade of baseball-reference), would be stats on multiple inning relief appearances. I believe that Doug Drinen compiled a lot of that data in his WPA listings for BBBA, and it's something that should be added to our knowledge base. I also suspect that Retrosheet could provide some insight into pre-1980 patterns.

Looking at the 1999 data, for example, it shows that Rivera was in 12 multiple-inning appearances (out of 66), while Trevor Hoffman was in 11 of 64.

I literally don't know if Torre is using Rivera in multiple-inning appearances at an astonishingly higher rate than the current MLB average, at least during the regular season. He clearly has done so in the post-season, which is why it's gotten so much attention.

Which leads me to another digression. Ken Adams and I were talking about complete game percentages in the post-season vs. the regular season and what we thought the relationship was. Our guess was that CG% was generally higher in the post-season, because better pitchers were used.

That turns out to be true for all of history, but not for the past twenty-five years (though there was a brief rally in the late 80s/early 90s due to the drop in offense during that time). I'll try to write a piece about this so that the chart with the amazing rise and dip in the 1965-76 time frame can be seen. My sense was that the A's World Series (1972-74) may have been the true dawning of the current usage patterns. There was not a single CG in those three Series, and the wide exposure of that approach may have been a defining moment.

The chart of the running 5-year WS CG% in that time frame literally looks like a man falling off a cliff...

To return on point, I think it would be useful if someone took play-by-play data and tried to examine how much difference these various managerial decisions actually make in the outcome of games. That's a helluva lot of work, but the way to cut it down to size would be to look just at games that fall into the "late and close" category. On a more general note, I think we'd all find it interesting to know how many games are "close" (within one or two runs) after five, six, seven innings as opposed to the final total of close games. Conversely, it'd be interesting to know how many games that aren't close at those points wind up being close games.

With that kind of information, we'd at least get a handle on how many games a manager's situational decisions really make a difference in the outcome.
   11. Don Malcolm Posted: November 19, 2001 at 12:14 AM (#604250)
?It's totally unfair to compare those save percentages. Obviously a multi-inning save is harder than the comparable one-inning save, because there are two chances to blow it. And on top of it, I'm guessing that multi-inning saves opportunities put the closer in tougher situations (usually one-run games, or runners on, or both). There are many easy one-inning saves, but I doubt that Mo Rivera sees any easy multi-inning saves.?

I suggest to you, Vinay, that under the current usage patterns there are precious few relievers who see easy multi-inning saves. The point wasn?t to compare the numbers and point a finger at Rivera, but simply to point out that he, like any reliever used in that way, will not be as successful in terms of converting those type of save opportunities.

STATS? ?easy-regular-tough? breakout for saves didn?t discriminate between single-inning and multiple-inning save opportunities, which was one of the reasons I didn?t quote any comparison data. Most of Rivera?s multi-inning saves, however, would fall into the ?regular? category, since Torre almost always brings him in to start an inning, which eliminates the ?tough? category (where the tying run must be on base).

For the 1996 season (last year of the STATS Scoreboard I have handy), the ?regular? save percentage (which includes, but is not solely encompassed by, multiple-inning appearances that do not involve the tying run being on base when the reliever enters the game) is 65%.

Multiple-inning saves almost certainly have a lower percentage than that, and Rivera?s record compared to other relievers is certainly going to be at or near the top of the list. The point was to show that there is some additional risk involved in using a reliever in that way, and that such a risk might be even greater when that type of usage becomes more concentrated (as is the case during the post-season).
   12. Kurt Posted: November 20, 2001 at 12:14 AM (#604257)
Don't forget that there are some easy multi-inning saves out there; namely, the "3 innings by the 10th pitcher up 10-2" type of saves. Obviously these are not a concern with looking at Rivera, since he doesn't get those saves, but when you're looking at the league as a whole they need to be accounted for since they're not germane to the discussion.

You must be Registered and Logged In to post comments.



<< Back to main

BBTF Partner

Dynasty League Baseball

Support BBTF


Thanks to
A triple short of the cycle
for his generous support.


You must be logged in to view your Bookmarks.


Page rendered in 0.4177 seconds
41 querie(s) executed