The Cardinals pulled out a huge 12 inning win (link to the Fangraphs play log) against the Los Angeles Dodgers on Sunday, which at the time gave them a one game lead for the second wild card spot. Using Fangraphs “Win Probability Added” stat I wanted to take a quick look at a few of outcomes of some of the small ball strategies that were used late in the game, and then provide an example or two of information that might be used when making a decision on using one of these strategies. I don’t intend this to be an evaluation of the managers and their decisions, as that is something I am definitely not qualified to do.

First, a quick note about WPA: it is a descriptive statistic. It is not meant to highlight a player’s true ability level, or be predictive about the future. Fangraph’s puts it well when they call it a “storytelling statistic.” It lets us quantify how each event in the course of a given game impacts a team’s probability of winning.

Intentional Walks:

  • Top of the 10th inning, Carlos Beltran was intentionally walked: +.015 WPA.
  • Bottom of the 10th inning, Juan Rivera was intentionally walked: -.003 WPA
  • Bottom of the 10th inning, Andre Ethier was intentionally walked: +.023 WPA
  • Top of the 12th inning, Carlos Beltran was intentionally walked: +.003 WPA.

A quick translation is that a + indicates an outcome that was net good for the offense, whereas a -WPA indicates is was good for the defense. As you can see, three of the four intentional walks were actually good outcomes for the hitting team. To begin to understand why, consider this quote from Tom Tango in The Book, “Issuing an IBB to an outstanding hitter with almost always increase the opposing team’s run production. The only situation this is a good idea is if the hitter is unprotected with men on second and third and two outs.” Ethier and Beltran have Matt Kemp and Matt Holliday hitting behind them, respectively, so I definitely wouldn’t say that either them are “unprotected” even if you want to grant that they meet the criteria of being an “outstanding” hitter.

This is a simplified analysis, and undoubtedly both managers were well aware of who was hitting next. To keep things brief, lets consider just the Dodgers’ decision to walk Beltran twice and dig a little deeper to see if we can identify some potential reasons for the decision. My first thought was that the Dodgers intended to take advantage of platoon splits by walking the switch-hitting Beltran to face the right-handed hitting Holliday. However, Holliday actually hits right-handed pitchers better than Beltran, posting a 139 wRC+ in his career against them, compared to just 125 for Beltran. Consider these next splits, however (courtesy of Fangraphs, all numbers are wRC+),

       Holliday         Beltran
Low Leverage

145

123

Medium Leverage

134

136

High Leverage

134

151

Bases Empty

141

118

Men on Base

136

137

Men in Scoring

122

133

These are their splits based on either the leverage of the situation or the position of runners on base over the course of their careers. The concept of clutch hitting is a complicated one. Tom Tango says, for example, “for all practical purposes, a player can be expected to hit equally well in the clutch, as he would be expected to do in an ordinary situation.” The folks over at Baseball Prospectus, however, suggest that some hitters may be 2-3% better in the clutch than we would expect, given their ordinary performance. With that in mind, I was surprised to see such big gaps between the high and low leverage performances of Beltran and Holliday over their careers, where the sample size should be large enough to regress those outcomes closer to their normal level of performance. Perhaps this factored into the Dodgers’ decision making. At the very least, this should give us an appreciation for the complexity of the decision that each manager faces.

Sacrifice Bunts:

  • Bottom of the 10th inning, A.J. Ellis sacrificed Mark Ellis from first to second with no outs: -.011 WPA
  • Top of the 12th inning, Pete Kozma sacrificed Matt Carpenter from first to second with no outs: -.019 WPA

I’ll only consider these two successful sacrifices by position players late in the game. Both were net bad for the offense, in terms of how the outcome of the play impacted the respective team’s chance of winning. This may catch some people off guard, as the bunt in the 12th inning by the Cardinals came in the inning where they scored the three runs that ultimately proved to be the difference in the game.

To begin to understand why these bunts were net negative. Consider the following table, again from the Baseball Prospectus team:

Situation      Goal          Ave         OBP          SLG          OPS
1st, 0 out 1+ Runs

0.191

0.206

0.182

0.388

1st, 1 out 1+ Runs

0.195

0.221

0.178

0.399

2nd, 0 out 1+ Runs

0.249

0.305

0.363

0.668

1st and 2nd, 0 out 1+ Runs

0.218

0.253

0.266

0.519

1st, 0 out 1 Run

0.177

0.192

0.153

0.345

1st, 1 out 1 Run

0.199

0.224

0.174

0.398

2nd, 0 out 1 Run

0.277

0.35

0.451

0.801

1st and 2nd, 0 out 1 Run

0.206

0.235

0.263

0.498

These numbers represent the threshold that the hitter must be below in order to make trading an out for a base a worthwhile exchange for the offense. I highlighted the line that is relevant for us here. In order for these bunts to have been considered a worthwhile trade for the offense, the hitter should have a sub-.345 OPS. Both A.J. Ellis of the Dodgers and Pete Kozma are well above this threshold (although Kozma has only played in 11 games so far). I believe that this chart takes into account all possible outcomes of a “sacrifice attempt” and doesn’t simply assume a successful sacrifice. The Baseball Prospectus folks argue that bunting a runner from first to second never increases a teams chance to score at least one run (or to have a big inning, for that matter). As you can see from the table, bunting a runner from second to third is a much more favorable outcome for the team that is simply trying to scratch a run across late in the game.

The Book also has some insights here. Tango looks more extensively at the decision to bunt a runner from first to second. He notes that the readiness of the defense for the bunt seems to have a big impact on the outcome of the play. The intuition is this: when a defense is expecting a bunt, they are much less likely to make an error, let it fall in for a single, or allow any of the other favorable outcomes for the offense. I feel comfortable saying that it didn’t look like either team was caught off guard by the bunt play in extra innings of a tie game. This matters, as simply advancing the runner to second isn’t a great outcome for the offense in this situation.

Wrapping Up:

The Baseball Prospectus team makes a disclaimer about the use of “win expectancy” (the underpinning of WPA) when evaluating managers. They note that they often don’t initiate the decision to steal, and sometimes a player bunts on his own. With that in mind they show that only 6 managers increased their team’s win expectancy from 1972-2004 by using sacrifices, steals, and IBB, reflecting that these plays are probably overused. Anyone that knows me is well aware that I tend to agree, but I’d love to know the process that goes into making these decisions at the big league level, as I’m sure it is much more detailed than my simple understanding presented here.

 

 

The information and quotes used here are taken from:

The Book: Playing the Percentages of Baseball, by Tom Tango, Mitchel Lichtman, and Andrew Dolphin

Baseball Between the Numbers: Why Everything You Know About the Game is Wrong, By the “The Baseball Prospectus Team of Experts” and Jonah Keri.