June 21, 2006
Lies, Damned Lies
The Other Molina
Why does anybody try and run on Yadier Molina?
Last season, only 39 runners tried to steal on him. In spite of their apparent selectivity, only 14 of those men succeeded. This was the fewest number of steals allowed by an everyday catcher since 1960 (minimum 100 games caught):
CATCHER YEAR SB CS G_C ----------------------------------------------------- Yadier Molina 2005 14 25 114 J.C. Martin 1964 15 17 119 Elston Howard 1962 19 23 129 Johnny Roseboro 1964 19 29 128 Elston Howard 1961 20 20 111 Elston Howard 1963 22 15 132 Gus Triandos 1961 22 20 114 Thurman Munson 1971 22 30 110 Bill Freehan 1964 23 26 141 Ivan Rodriguez 2001 23 35 106 Paul Casanova 1970 23 24 100Now, 2005 was Molina's first full season in the league, and opponents were having a tough enough time scoring runs against the Cardinals, so perhaps their aggression can be excused. But already this season, 23 men have tried to steal in Molina's 51 games behind the plate. True to form, he's gunned down 11 of them. We can debate whether the breakeven success rate for a stolen base attempt is 67 percent or 75 percent or something else, but surely even the Dusty Bakers of the world can understand that creating an out half the time you attempt to steal a bag simply isn't worth it.
Well, perhaps not. These are Ivan Rodriguez' opposing basestealer numbers over the course of his career. SBR stands for stolen base runs, a simple linear formula that assigns .2 runs for a stolen base and deducts .5 runs for a time caught stealing (we'll get to the rationale for these numbers in a moment).
Catcher Year SB CS SBR ------------------------------------------------------------ Ivan Rodriguez 1991 36 34 -9.8 Ivan Rodriguez 1992 53 57 -17.9 Ivan Rodriguez 1993 64 51 -12.7 Ivan Rodriguez 1994 37 23 -4.1 Ivan Rodriguez 1995 40 37 -10.5 Ivan Rodriguez 1996 46 48 -14.8 Ivan Rodriguez 1997 37 49 -17.1 Ivan Rodriguez 1998 38 49 -16.9 Ivan Rodriguez 1999 34 41 -13.7 Ivan Rodriguez 2000 20 19 -5.5 Ivan Rodriguez 2001 23 35 -12.9 Ivan Rodriguez 2002 26 15 -2.3 Ivan Rodriguez 2003 40 20 -2.0 Ivan Rodriguez 2004 40 19 -1.5 Ivan Rodriguez 2005 33 3 -10.9 Ivan Rodriguez 2006 9 10 -3.2Year after year after year, Rodriguez' opponents have given up ten runs or more in attempting to steal on him, in spite of his long history as a Golden Gun. I hope you can see what I'm getting at here. Why do managers so consistently elect a strategy that so clearly has a negative expected return? If a catcher throws out runners at something much greater than the breakeven rate, shouldn't the steal attempts against him quickly dwindle to zero?
Well, there are a couple of reasons that it might not, even if managers were applying perfect strategy. What Keith Woolner's research in Baseball Prospectus 2006 reveals is that the average breakeven success rate over the past ten seasons has been just about 71%. This is how the SBR formula is derived; coefficients of .2 and .5 imply a breakeven success rate of .714. However, the breakeven rate varies from situation to situation. If you have a runner on first base in a tied game in the bottom of the ninth, with two outs and a singles hitter at the plate, a stolen base is tremendously valuable; perhaps you could stomach a 50% success rate, or even a bit less. Other stolen base attempts are a lot more gratuitous. If the opponents of a Pudge or a Yadier Molina steal only when they can get away with a 50% success rate, then we shouldn't be surprised if their results reflect this.
The other reason is that a lot of stolen base attempts aren't really straight steals. If the runner is leading off first base too much, draws a pick-off throw, and breaks for second, that is recorded as a stolen base attempt. If a runner is breaking for home on a squeeze play, and the batter misses the pitch completely, that goes as a steal attempt, too. A more frequent occurrence is a steal attempt that results from a hit-and-run play where the batter fails to put the ball into play.
I went back and looked at play-by-play logs (and in some cases, MLB.tv video) for the 23 steals that have been attempted against Yadier Molina thus far on the season. I classified a steal attempt as H&R (hit-and-run) if it occurred with two strikes, or if the batter swung at and missed the pitch.
Date Opponent Inning Score Runners Outs Count Result Type 8-Apr Derrek Lee 2nd Tie 1st, 3rd 1 3-2 SB H&R 14-Apr Felipe Lopez 6th +1 1st 0 1-2 SB H&R* 18-Apr Jason Bay 4th +7 1st 2 1-1 SB Straight 22-Apr Ronny Cedeno 5th +1 1st 1 1-0 SB Straight 24-Apr Jason Bay 1st Tie 1st 2 3-2 CS H&R 27-Apr Alfonso Soriano 1st Tie 1st 0 1-2 CS H&R 2-May E. Encarnacion 8th Tie 1st 2 3-2 CS Pickoff 6-May Miguel Olivo 2nd -1 1st 1 0-2 SB H&R 7-May Joe Borchard 2nd -2 1st 0 3-1 CS Straight 9-May Jamey Carroll 1st Tie 1st 2 2-2 SB H&R 12-May Damion Easley 1st Tie 1st 2 1-0 CS Straight 12-May Orlando Hudson 5th -2 2nd 1 0-2 SB H&R 14-May Craig Counsell 5th +1 1st 1 1-0 CS H&R* 24-May Omar Vizquel 3rd Tie 1st, 2nd 0 2-0 SB Straight 27-May Dave Roberts 7th -1 1st 0 1-0 CS Straight 28-May Mike Cameron 1st Tie 1st, 2nd 1 3-1 SB H&R* 28-May Josh Barfield 8th +3 1st 2 1-0 SB H&R* 2-Jun Tony Womack 1st Tie 1st 1 0-1 CS Straight 2-Jun Ronny Cedeno 7th +1 1st, 3rd 1 1-0 CS Sac** 2-Jun Freddy Bynum 10th Tie 1st 2 2-0 CS Straight 3-Jun Juan Pierre 4th +4 1st 2 2-1 SB H&R* 7-Jun Felipe Lopez 5th +2 1st 0 2-1 SB Straight 16-Jun Jamey Carroll 3rd Tie 1st 2 3-1 CS Straight * Swinging strike ** Attempt at squeeze play; picked offAs you can see, of the 23 steal attempts recorded against Molina thus far on the season, 11 were clear hit-and-run plays, one was a pick-off attempt where the runner broke for second, and another one was a botched squeeze play. That means that at most 11 were straight steals. I say "at most" because I did not review the video for all of these, and it is possible that several others could have been hit-and-run attempts on counts where the batter still had a strike left to burn, the result of a pickoff attempt, and so forth. A couple of others that I did review on video were unusual plays. Omar Vizquel's steal of third on May 24th, for example, was a steal on pitcher Tyler Johnson; Vizquel was most of the way to third base by the time the pitch was thrown. It's only about once a week or so that managers are calling a straight steal against Molina, which is about as close to shutting down the running game completely as I think we're liable to witness.
The complication is that, in general, it would seem that teams are giving up runs with the stolen base. As I described earlier, the breakeven success rate over the past ten years or so is about 71%. However, the actual success rate during this period is 69%. So perhaps having a catcher like Molina isn't so valuable at all, since he only serves to deter a strategy that opponents apply incorrectly to begin with.
The first thing we should ask is: what should the actual steal success rate be, assuming that both the offense and the defense are applying perfect strategy? At first glance, it would seem that it should be a fair bit higher than 71%, because a manager should only attempt a steal when he has at least a 71% chance of success, and sometimes situations will arise--Carl Crawford on first with Mike Piazza behind the plate--when the implied success rate is clearly higher than 71%. However, the quality of the baserunner and the catcher aren't the only factors impacting steal success rates. In particular, the defense can adopt counter-measures, such using pick-off attempts, calling for a pitchout, or playing the second baseman closer to the bag, that reduce the offense's success rate (or deter it from stealing in the first place).
The problem is that these counter-measures have consequences that are undesirable for the defense. A pitchout gets the pitcher into a worse count, a non-optimal defensive alignment will improve batting averages, and a pickoff can result in an error, tire out the pitcher, or simply get the defense out of its rhythm. Note also that these negative consequences are the result of the threat of a stolen base, not a stolen base itself. Carl Crawford steals a lot of bags, but he also induces a lot of pitchouts, which is helpful to the batters hitting behind him. On the other hand, a manager that only stole in 'perfect' situations would not benefit from these externalities; there is therefore some incentive to steal a bit more often than the particular situation might seem to require.
Although my brain is a little bit too fuzzy at the moment to work through all the game theory permutations, I suspect that the equilibrium might well be an actual stolen base success rate that is very close to the idealized success rate of 71%. But this does not mean that the offense would score as many runs (or more runs) if the stolen base were not a legal play. In exchange for getting actual stolen base success rates back down to the breakeven mark, the defense pays a price in the form of extra pitchouts and so forth.
However, this price is less if the catcher has a strong throwing arm. These are the number of pitchouts that Tony LaRussa has called for since taking over as the manager of the Cardinals in 1996:
Year Primary C Pitchouts -------------------------------- 1996 Pagnozzi 41 1997 Difelice 79 1998 Marrero 34 1999 Marrero 30 2000 Matheny 34 2001 Matheny 25 2002 Matheny 13 2003 Matheny 9 2004 Matheny 7 2005 Molina 9With average defensive catchers like Mike Difelice and Eli Marrero behind the plate, LaRussa was pickoff-happy, calling for as many as 79 pitchouts in a season. At some point during the 2001 or 2002 season, however, LaRussa recognized that having a great defensive catcher like Molina or Mike Matheny behind the plate eliminated the need to call for pitchouts. The runner probably wasn't going to bother to steal anyway--and if he did, he might well be thrown out. Alan Trammell, in Detroit, realized the same thing. In 2003, with Brandon Inge as his primary catcher, he called for 28 pitchouts. The next year, the Tigers acquired Ivan Rodriguez, and Trammell pitched out only 8 times.
Therefore, I suspect that the benefits of a catcher like Molina are understated by conventional sabermetrics. Molina's contribution is not so much in eliminating the stolen base, but in eliminating the threat of one.
Acknowledgement: this column was somewhat inspired by The Book, which is recommended to all my regular readers.