To Go or Not to Go?

“Baseball is a kind of collective chess with arms and legs in full play under sunlight.” – Jacques Barzun from God’s Country and Mine

Recently I was looking through some data as a favor for our friends over at the excellent site Nats Blog, and I ran into a few interesting tidbits I thought I’d trot out this week.

The topic is sacrifice flies, so this also provides a segue with John Walsh’s recent articles “Do Batters Try to Hit Sacrifice Flies?” and “Can Batters Successfully Modify Their Batting Approach?” as well as my recent articles “Not So Sweet Surrender” and “Sacrificing in 2005 Redux.”

In a nutshell, what I was looking at was how the success rate changes for sacrifice flies when the batter hits the ball to the three outfield positions. In other words, how much more likely is a runner to tag and score when the fly ball is caught by a left or right fielder as opposed to a center fielder?

To Go or Not to Go

To look at this question I used the 2003-2005 play-by-play data and examined the results of all 4,950 fly balls that were hit with a runner on third and fewer than two outs that did not result in either a hit or a dropped-ball error on the outfielder. (This includes throwing errors.) Breaking the results down by the position that caught the fly ball you find the following:

Position           Opp   Hold   Hold% Scores   DP    DP%   Succ%
LF                1467    293   0.200   1119   55  0.037   0.953
CF                1910    225   0.118   1632   53  0.028   0.969
RF                1573    318   0.202   1197   58  0.037   0.954
                  4950    836   0.169   3948  166  0.034   0.960

Here, the Opportunities are the number of fly balls, Hold% represents the percentage of time the runner stayed on third, and the Succ% the percentage of times the runner scored when sent.

So when a fly ball is hit to the center fielder, the runner tags about 9% more often than when the ball is caught by either of the corner outfielders. Runners are also more successful by about 1.5% when the ball is in the hands of the center fielder. Overall, though, when a runner is sent, he makes it more than 19 times out of 20 regardless of who catches it.

A conclusion one might draw from this is that third base coaches are more cautious when the ball is hit to the corners both because some corner outfielders (Ichiro Suzuki comes to mind) have better arms but probably more likely because the ball is simply closer to the plate when caught by a corner outfielder. Alas I don’t have distance data to confirm just what the mix of these two factors might be.

So far so good and pretty much what we’d expect.

What surprised me a little was when I looked at the data broken down by the number of outs.

Outs               Opp   Hold   Hold% Scores   DP    DP%   Succ%
0                 1387    257   0.185   1097   33  0.024   0.971
1                 3563    579   0.163   2851  133  0.037   0.955

Here what you’ll notice is that runners were sent from third 81.5% of the time (1-.185) with nobody out and 83.7% of the time with one out. While this difference moves in the direction you’d expect, with more runners being sent and more being caught with one out than with nobody out, what surprised me is that the difference was not larger, since “the book” would tell you that it is a far worse error to get a runner thrown out as the second out of an inning than as the third.

This conclusion is easy to see when you apply Run Expectancy and Scoring Probability matrices to the question. Yes, I realize that these matrices are not that useful when analyzing very specific situations, because the values are affected by the context which includes the batter, pitcher, fielders, score, lineup, park, weather and what the umpire had to eat before the game. But here we’re talking about the general application of a strategy using aggregate data, so I think we’re on more solid ground.

In any case, using the Run Expectancy and Scoring Probability tables you can access on Tangotiger’s site and the break-even formula I described in my first article on bunting, you can make some simple estimates as to when it makes sense for a third base coach to send his runner. For example, when there is a runner only on third with nobody out, the calculation indicates that the coach should only send the runner if he’s certain the runner will make it when the goal is to maximize runs. However, if his aim is to score a single run (i.e. when that run will tie or win the game), the coach should send the runner if he thinks the runner’s odds of making it are greater than around 85%. With one out the needed probability of success drops to 88% and 66% respectively.

Of course the calculation also changes when there are other runners on base. For example, with runners on first and third and one out, the needed probability of success to maximize runs increases to 92.4%.

Because of the relatively wide difference between these strategic tipping points and the data indicating that runners are actually successful 95-97% of the time, I thought that perhaps the Hold% would change when examining the data by score differential and reveal that indeed third base coaches took more risks when down by a run or with the game tied.

Score Diff         Opp   Hold   Hold% Scores   DP    DP%   Succ%
>=-5               353     70   0.198    274    9  0.025   0.968
-4                 188     25   0.133    161    2  0.011   0.988
-3                 273     43   0.158    225    5  0.018   0.978
-2                 396     83   0.210    298   15  0.038   0.952
-1                 576    102   0.177    447   27  0.047   0.943
0                 1131    186   0.164    904   41  0.036   0.957
1                  639    103   0.161    518   18  0.028   0.966
2                  458     78   0.170    360   20  0.044   0.947
3                  320     53   0.166    258    9  0.028   0.966
4                  239     31   0.130    198   10  0.042   0.952
>=5                377     62   0.164    305   10  0.027   0.968

As you can see, when looking at the Hold% there really is no discernable pattern that would indicate that third base coaches are more aggressive in some situations than others, although when down by a run the Success% but not the Hold% does drops a bit to 94.3%.

A Hardball Times Update
Goodbye for now.

To summarize the point, when you compare the needed scoring probability using the matrices with what actually happens, it would appear that third base coaches only send runners when they are virtually certain they’ll make it. We can infer this because if they took more risks we would likely see lower success rates. The point becomes even stronger when you recall that errors aren’t included in these statistics. This behavior, from a strategic perspective anyway, is suboptimal. Chalk it up to the human propensity to be risk averse.

Intimidation

As mentioned previously, one reason third base coaches might be more reticent about sending the runner is related to the strength of the arms in the outfield. Because of this, as many baseball fans are well aware, the mark of a great outfield arm is usually not how many assists a player racks up but rather how few times an opposing team challenges his arm.

This axiom is particularly applicable to sacrifice fly opportunities, so I also took a quick look at the data broken down by the fielder who caught the fly ball. The following table shows the top 20 corner outfielders with 15 or more opportunities sorted by Hold%. In other words, this might give us an indication of which arms are most feared.

Name            POS        Opp    Hold Hold% Scores     DP     DP%   Succ%
Brad Wilkerson  LF          20       9 0.450      8      3   0.150   0.727
Craig Monroe    RF          24      10 0.417     13      1   0.042   0.929
Danny Bautista  RF          23       9 0.391     14      0   0.000   1.000
Ichiro Suzuki   RF          59      22 0.373     33      4   0.068   0.892
Raul Mondesi    RF          19       7 0.368     11      1   0.053   0.917
Jose Guillen    RF          25       9 0.360     15      1   0.040   0.938
Jose Guillen    LF          17       6 0.353     10      1   0.059   0.909
Luis Gonzalez   LF          32      11 0.344     20      1   0.031   0.952
Geoff Jenkins   RF          15       5 0.333      7      3   0.200   0.700
Terrence Long   LF          16       5 0.313     11      0   0.000   1.000
Matt Lawton     LF          16       5 0.313     11      0   0.000   1.000
Trot Nixon      RF          33      10 0.303     23      0   0.000   1.000
Manny Ramirez   LF          41      12 0.293     29      0   0.000   1.000
Raul Ibanez     LF          41      12 0.293     26      3   0.073   0.897
Jacque Jones    RF          24       7 0.292     17      0   0.000   1.000
Geoff Jenkins   LF          21       6 0.286     13      2   0.095   0.867
Jeromy Burnitz  RF          25       7 0.280     18      0   0.000   1.000
Juan EncarnacionRF          43      12 0.279     30      1   0.023   0.968
Kevin Mench     LF          18       5 0.278     12      1   0.056   0.923
Brian Giles     RF          29       8 0.276     19      2   0.069   0.905

This list tracks fairly well with the word on the street since Brad Wilkerson, Danny Bautista, Ichiro Suzuki, Raul Mondesi and Jose Guillen are all reported to have strong arms. On the other side of the coin Austin Kearns and Reed Johnson have the distinction of never having a coach hold his runner in 24 and 16 opportunities respectively. Other “poor arms” noted here were Jason Bay (one hold in 29 opp), Brady Clark (2-27), Shannon Stewart (2-27), Rondell White (3-30) and Carl Crawford (5-46).

And here are the top center fielders.

Name            POS        Opp    Hold Hold% Scores     DP     DP%   Succ%
Torii Hunter    CF          40      12 0.300     26      2   0.050   0.929
Jay Payton      CF          25       6 0.240     18      1   0.040   0.947
Kenny Lofton    CF          38       9 0.237     27      2   0.053   0.931
Wily Mo Pena    CF          19       4 0.211     15      0   0.000   1.000
Luis Matos      CF          41       8 0.195     31      2   0.049   0.939
Gary Matthews   CF          31       6 0.194     25      0   0.000   1.000
Mike Cameron    CF          38       7 0.184     31      0   0.000   1.000
Scott Podsednik CF          40       7 0.175     33      0   0.000   1.000
Aaron Rowand    CF          41       7 0.171     34      0   0.000   1.000
Andruw Jones    CF          53       9 0.170     42      2   0.038   0.955
Brady Clark     CF          18       3 0.167     15      0   0.000   1.000
Milton Bradley  CF          37       6 0.162     29      2   0.054   0.935
David DeJesus   CF          31       5 0.161     26      0   0.000   1.000
Jim Edmonds     CF          56       9 0.161     44      3   0.054   0.936
Alex Sanchez    CF          33       5 0.152     28      0   0.000   1.000
Rocco Baldelli  CF          47       7 0.149     37      3   0.064   0.925
Brad Wilkerson  CF          27       4 0.148     22      1   0.037   0.957
Carlos Beltran  CF          69       9 0.130     59      1   0.014   0.983
Marlon Byrd     CF          18       2 0.111     16      0   0.000   1.000
Endy Chavez     CF          18       2 0.111     14      2   0.111   0.875

On the bottom of the pile we find Marquis Grissom and Bernie Williams who were run against in every one of their 38 and 43 opportunities. Nook Logon (1-25), Randy Winn (1-24) and Jeremy Reed (1-20) also bring up the rear.

It should be noted that park effects aren’t taken into account here, so left fielders like Manny Ramirez and Moises Alou may be benefiting from a small left field area, while Hideki Matsui is likely being penalized. In addition, because we don’t have distance data, there are some outfielders who were probably unlucky in the sense that they happened to catch more than their share of long fly balls in these situations when they have no chance of throwing out the runner.

The fielders can also be ranked by their ability to throw out the runner, so here are the top 10 corner outfielders in Succ%:

Name            POS        Opp    Hold Hold% Scores     DP     DP%   Succ%
Geoff Jenkins   RF          15       5 0.333      7      3   0.200   0.700
Brad Wilkerson  LF          20       9 0.450      8      3   0.150   0.727
Miguel Cabrera  LF          28       4 0.143     20      4   0.143   0.833
Richard Hidalgo RF          38      10 0.263     24      4   0.105   0.857
Alexis Rios     RF          30       8 0.267     19      3   0.100   0.864
Geoff Jenkins   LF          21       6 0.286     13      2   0.095   0.867
Ichiro Suzuki   RF          59      22 0.373     33      4   0.068   0.892
Eric Byrnes     LF          24       5 0.208     17      2   0.083   0.895
Raul Ibanez     LF          41      12 0.293     26      3   0.073   0.897
Brian Giles     RF          29       8 0.276     19      2   0.069   0.905

And the center fielders:

Name           POS        Opp    Hold Hold% Scores     DP     DP%   Succ%
Endy Chavez    CF          18       2 0.111     14      2   0.111   0.875
Randy Winn     CF          24       1 0.042     21      2   0.083   0.913
Mark Kotsay    CF          54       5 0.093     45      4   0.074   0.918
Rocco Baldelli CF          47       7 0.149     37      3   0.064   0.925
Steve Finley   CF          43       3 0.070     37      3   0.070   0.925
Torii Hunter   CF          40      12 0.300     26      2   0.050   0.929
Kenny Lofton   CF          38       9 0.237     27      2   0.053   0.931
Milton Bradley CF          37       6 0.162     29      2   0.054   0.935
Jim Edmonds    CF          56       9 0.161     44      3   0.054   0.936
Luis Matos     CF          41       8 0.195     31      2   0.049   0.939

Of course small sample size warnings apply to these lists since it only takes a couple of good throws to drive down the Succ%.


Comments are closed.