And that ain’t all, he stole home!by Shane Tourtellotte
March 02, 2012
It is our most enduring image of Jackie Robinson. A dance off third base, a break for home, the slide for the plate and Yogi Berra's glove and—SAFE!—and not even acknowledging Yogi's rhubarb with the umpire as he picks himself up and circles away. Robinson's steal of home in Game One of the 1955 World Series marked him forever as the daring, almost reckless runner who could and would steal any open base ahead of him, especially home plate.
The image may simplify, but it doesn't lie. His penchant for stealing home was strong enough that it made its way into the 1949 song "Did You See Jackie Robinson Hit That Ball?," whose lyrics I borrowed for my title. (If you don't know the song, go here and have a listen. It's three minutes. I'll wait until you're back.) The fateful meeting with Yogi merely gave us the visual hook upon which to hang an established reputation.
Reputation, however, is not enough for our generation of baseball fans. We know base-running aggression can quickly turn counter-productive. We know spectacle can be hollow. Did Jackie Robinson's dashes for home bolster the Brooklyn Dodgers, or only his image? I decided to look at that question, and some related ones. My survey produced plenty of information, including one discovery that I confess had me flabbergasted when I first ran the numbers.
Three notes before I proceed. First, the bulk of my data come via Retrosheet, beginning at its tribute page to Robinson filled with statistics drawn from the personal score sheets of Allan Roth, ur-sabermetrician for the Dodgers. It is with gratitude that I deliver the following familiar incantation:
The information used here was obtained free of charge from and is copyrighted by Retrosheet. Interested parties may contact Retrosheet at "www.retrosheet.org."
Second, the data for the first few years of Jackie Robinson's career are not as comprehensive as we are used to for more modern seasons. Expected Runs tables (courtesy of Baseball Prospectus) went back only to 1950 when I did my research, leaving Robinson's three most productive years of stealing home in a fuzzy limbo for an analyst. I have been forced as a workaround to calculate those three years' steals by the 1950 Expected Runs matrix, the closest year available.
This very likely skews the numbers, particularly since it was the highest-offense year of the 1950-1956 span, quite possibly a one-year fluke, and the break-even numbers on steals were commensurately higher. Robinson may be unfairly penalized for 1947-1949, his most prolific home-stealing years. For the same reasons, I must eschew Expected Wins, even though it would be more precise in figuring the value of his plays. When further data become available, however, I intend to revise this article to reflect our new knowledge.
Third, the subject by its nature bears the burden of small sample size. You are invited to take the proverbial grain of salt, or maybe a larger dose, now, because I am going to bull ahead and draw conclusions from the data without burying myself in error bars or regressions to the mean. It's not like I have to worry about making an inaccurate projection of Robinson's stealing prowess for the upcoming year. His seasons are long past. I will take them for what they were.
Now, the simple numbers. Jackie Robinson stole home 19 times in his regular season career, and was caught stealing home 12 times, for a rate of 61.3 percent. He had his famous steal against Whitey and Yogi as his lone World Series attempt, nudging the numbers to 20 for 32, 62.5 percent. I think it's fair to include this in his lifetime total; I would have done likewise had he gotten caught. (And now the sample is a little larger. Feel better now?) The specifics of each attempt are listed in this table.
Scores always show Brooklyn first. Underlined numbers under "Base/Out" signify runners who were stealing along with Robinson on the play. Underlined "ΔER" numbers are ones I adjusted to reflect something other than the raw change in Expected Runs, twice due to busted plays (a hit-and-run on 7/26/47 and a suicide squeeze on 5/13/56), and twice due to other base-stealers attempting to take multiple bases, for which Robinson cannot be fairly credited (7/4/48) or debited (8/29/55).
That steal percentage of 62.5 percent brings a quiver to the heart of people who know the numbers for regular steal attempts. Anything below 70 percent we automatically categorize as counter-productive, subtracting runs, hurting the team. But going for home is a different animal than stealing second or even third, and the break-even numbers do reflect this.
The best situation for stealing home is with a runner on third only and two outs. During Robinson's career, the break-even rate ranged from a low of 33.7 percent to a high of 38.5 percent. This is a somewhat more unfavorable environment than today: From 2009 to 2011, the rates have been 33.5/31.0/32.3. But suddenly Robinson's 62.5 percent success rate looks awfully good—if one assumes he was going for home only in optimal situations.
He wasn't. Only 15 of his 32 career attempts were in XX3/2 situations, with an eight for 15 success rate. Other attempts came with men on base, or with only one out, or both, and the break-even numbers rose accordingly. Seven of his attempts had break-evens over 50 percent—all with one out—and three managed to exceed 70 percent, two with another runner on base who was not also stealing. (This lowers the Expected Runs gain on a success, as an additional advance always boosts the ER.)
One of the 70+ affairs was his very first steal of home on June 24, 1947, and one can argue he had been goaded into some payback, as Pirates hurler Fritz Ostermueller had earlier thrown at Robinson's head. But Robinson could buck the odds without such motivation—not that he likely knew the odds. Sabermetrics was in its infancy, and Allan Roth was only one man, a man without even a primitive computer.
But we know the odds, at least for seven of Robinson's 10 years. Given this proviso, however much it raises your sodium intake, the numbers say Robinson's attempted steals of home had a mean break-even rate of 46.1 percent. This comes to 14.76 steals in 32 attempts, and Robinson made 20. Even with the more reckless tries thrown in, Robinson was well into positive territory with his steals of home.
The gain in Expected Runs comes out as +8.81, close to one full win over his career, going by the standard 10 runs = 1 win rule of thumb. Whether this seems low or high depends on whether you think of it as his whole career, or as the result of a mere 32 events. Once we have full Expected Wins data, though, it might end up less than we might anticipate.
Robinson may be penalized on Expected Runs by the lack of information we have for 1947-1949, but a countervailing factor is the environment in which he did his stealing. The Dodgers were a strong offensive team during his tenure, leading the National League in OPS+ seven of 10 years. The potent bats around him, plus an offense-friendly Ebbets Field as home turf, altered the run environment and raised the break-even levels on steals from the league average. My Expected Runs totals are thus a little inflated from his true value—which may approximately cancel out the aforementioned debit. Another grain of salt.
I will depart for a while from deep-geek calculations to examine what could fairly be called the trivia of Robinson's home-stealing career, which produce some interesting nuggets of their own.
The majority of Jackie's attempts on home came early in his career. He made 19 attempts in his first three seasons—3-1 in '47, 5-3 in '48, 5-2 in '49—and then he never again had more than two regular-season attempts in a year. There are two likely explanations: he was getting older and slower (though the drop-off from 1949 to 1950 seems severe), and pitchers started expecting him to try something funny and thus were on guard, reducing his opportunities. (I guess they heard the song.) Both are probably true to some extent, but there may be a third explanation.
In 1947 and 1948, the Dodgers were offensively rather average, third in league OPS+ in '47 and tied for fifth in '48. Robinson's big-league career began in this environment, and surely informed his base-stealing decisions to some extent. In 1949, the bats began booming, starting Brooklyn's seven-year string of OPS+ titles.
I suggest that, after experiencing the first year of this power surge, Robinson realized he didn't need to be so aggressive in manufacturing runs, and curtailed his steals of home accordingly. By the time Brooklyn's offensive potency waned, Robinson was in his final season, a grizzled 37, and knew better than to tempt fate. I make no guarantees for this theory of mine, but for a player as intelligent as Robinson was, it seems a reasonable complement to the more obvious ones.
Robinson made no special effort to steal against left-handed pitchers, who in a reversal of the rule for stealing second are easier to beat on steals of home than righties. His first three career attempts were against southpaws—perhaps still feeling his way?—but lifetime he went 10 times against lefties and 22 times against RHPs. LHPs accounted for roughly 30 percent of batters faced in this time frame, so Robinson's 31 percent rate shows no preference for testing the portsiders.
He did seem more comfortable stealing home when he was at home. The home/road splits for his attempts are 19/13. His favorite road site for the play was Pittsburgh, with five attempts at Forbes Field, four in his first two seasons.
A dash home by Jackie was usually a good omen for the Dodgers: They went 23-9 in games where he attempted to steal home. However, this almost perfectly reflects the game situations in which he stole. Of his 32 attempts, 20 came with the Dodgers ahead, seven behind, and five tied. Split those ties 3-2 for Brooklyn, reasonable for a consistently winning team, and it would come out an identical 23-9. Success or failure seemed to make no difference: The Dodgers were 14-6 when he stole home, 9-3 when he was caught.
This reflects a larger truth about Jackie's stealing: He did not tailor it to high-leverage situations. Of the 13 lifetime events for which we have Leverage Index data, the average LI of his steal attempts was 1.29 (the baseline being 1). This sounds elevated, but consider the context: There's a runner on third, maybe more, and usually two outs, and those facts naturally raise the leverage. In Dan Turkenkopf's survey of steals of home in the 2000s, he found an average LI of 1.8 (from an admittedly small sample size of 25), against a 1.2 average LI of steals of any base.
The rawer, less analytic numbers back this up. Only two of his 32 attempts came in late and close situations (one in the Series), half or less what we'd expect from random chance. Eight attempts came with a margin of at least four runs, including the day in Boston when he stole on the Braves with a 10-run lead in the eighth! (Manager Charlie Dressen promptly substituted for Robinson, presumably knowing he had made himself a prime target for a retaliatory plunking. The Braves had to settle for hitting Carl Furillo.)
This is another reason why, when we have full Expected Wins data for Robinson's career, his steals may end up a little less valuable than they currently appear. If anything, Robinson was more likely to go at less pivotal moments possibly because pitchers weren't as vigilant, possibly because Robinson regarded the steal of home as less a tactic than a psychological weapon, but those suppositions are beyond the reach of statistics.
This was not the astonishing conclusion I promised you earlier. For that, we need to examine half a dozen steal attempts that I maintain were unrepresentative of Robinson's skills. Two I mentioned earlier: the failed hit-and-run and suicide squeeze that left Robinson hung out to dry (though he made a nice long rundown out of the hit-and-run). The four others were delayed double-steals: runner on first breaking, and Robinson going for home on the throw toward second.* Said throw would presumably be cut off to nail Robinson at the plate, which usually happened: He was only one for four on those attempts.
* Dates were 7/19/47; 9/3/48; 8/14/49; 5/2/51.
For a moment, I want to act as if those six plays didn't happen.
Exclude the two whiffs on the hit-and-run and suicide squeeze—those were the batters' failures, not Robinson's. Also exclude the four delayed double-steals, where Robinson was dependent on how the defense reacted to another baserunner for his success. Take the remaining 26 attempts, where it was all Robinson's responsibility to get the right lead, watch for the pitcher's moment of inattention or the hitch in his motion, and choose the perfect instant to break for home.
In those attempts, Robinson was 19 for 26 lifetime, a 73.1 percent success rate. His career record for steals other than home was 182 for 246*, a success rate of 74.0 percent. They are, amazingly, almost equal. Even were we to comb the busted hit-and-runs out of the career totals and raise the rate a point or so, we would still be left with an astonishing conclusion. When left to his own devices, Jackie Robinson took the most dangerous play in baseball and made it virtually no likelier to produce an out than an everyday steal of second.
* This is taken from Allan Roth's numbers at Retrosheet, which credit Robinson with one fewer stolen base than the official records.
I said I was flabbergasted, and I stand by that statement.
The overall steal numbers for Robinson nudge me toward a second conclusion. That 74.0 percent is only a fairly good rate: it beats the yearly break-even percentages by slender margins. I would have to study all of his steal attempts individually to gauge it properly, a fairly daunting task, and the missing ER and EW data from the 1940s makes it impossible for now to get a full analytic grip on the matter. But once those numbers are in, and for anyone willing to do the work—which may include me at that later date—I offer the following.
Hypothesis: Jackie Robinson accumulated more run-producing, game-winning value with his steals of home than with all his other career steal attempts put together.
That's a conclusion, or a disproof, for another day, but we have enough findings for now. Jackie Robinson's skill at stealing home has not been exaggerated by his reputation. If anything, it's the rest of his stealing game that gets inflated. He produced better than a quarter-run of ER per attempt, a rate he could not have reached stealing second had he been perfect at it.
The timing of his attempts was never the sabermetric ideal, but nobody knew the ideal then, and Robinson was frankly too intense a competitor to be a push-button ballplayer. Precisely how much his swipes of home benefited the Dodgers is beyond our reckoning, but I hope not for long as full records are developed for his early years in Brooklyn. There is more to be mined here, and Jackie Robinson's career will benefit from further study.
That, without doubt, is the least unexpected finding in this entire piece.
References and Resources
Online archives of The New York Times, various dates 1947-1956
Jonathan Eig, Opening Day
Shane Tourtellotte is a long-time, occasionally-nominated science fiction writer, currently living in Asheville, North Carolina. He will tell you all about the baseball novel he’s shopping if you give him an inch.