/cdn.vox-cdn.com/uploads/chorus_image/image/40894742/20140708_jla_sv7_739.jpg.0.jpg)
The narrative for the Royals this postseason is small ball, a term used to denote teams that use speed and sacrifice bunts to score runs. While it's familiar to those with knowledge of baseball history, it's a strategy that fell by the wayside with the introduction of more potent offenses in the mid-1990s. The term was revived briefly to describe the 2005 White Sox, a team that was near the top of the AL in stolen bases, led in sacrifice hits, but also hit 200 home runs -- small ball with a very healthy dose of slugger ball.
The Royals lived up to the moniker this season -- they were last in the AL in home runs and walks by a fairly healthy margin, but led in stolen bases. They had five players with double-digit steals and presented a fresh alternative for those who yearn for old-fashioned baseball.
There's just one problem -- one of the pesky side effects of the increased gathering and processing of baseball information is the ability to determine what works and what doesn't. In The Book, Tom Tango, Mitchel Lichtman and Andrew Dolphin developed a very useful tool to determine the effectiveness of almost any strategy, the Run Expectancy Matrix. This chart shows the average number of runs scored by base/out situation until the end of an inning:
Used with Tom Tango's permission
This brief example explains how to view the chart: With a runner on first and no outs, teams can be expected to score .941 runs (in all cases I'll use the 1993-2010 table). If he were to steal second, the expectation would be teams would score 1.170 runs -- the stolen base increased the expected runs by .229 runs. However, if he's thrown out trying to steal, the expected runs for that runner drop to zero. In other words, the added value of the stolen base (.229 runs) has to be balanced against the risks of being thrown out (-.941 runs). In those terms, being thrown out has four times the cost of the value of stealing second; therefore, teams better be successful in their stolen base attempts about four times as often as they're thrown out.
These are the Royals' stolen bases through Game 2 of the ALDS viewed through the Run Expectancy Matrix expected values:
G | Inn | Score | Outs | RoB | Event | RE | Success | Diff | Fail | Diff |
---|---|---|---|---|---|---|---|---|---|---|
WC | 1 | 0-2 | 2 | 1-- | Aoki Steals 2B | 0.245 | 0.348 | 0.103 | 0 | -0.245 |
WC | 1 | 1-2 | 2 | 1-3 | Hosmer Caught Stealing Hm (P-SS-1B-C) | 0.53 | 1.245* | 0 | -0.53 | |
WC | 8 | 3-7 | 0 | 1-- | Escobar Steals 2B | 0.941 | 1.17 | 0.229 | 0 | -0.941 |
WC | 8 | 4-7 | 1 | 1-- | Cain Steals 2B | 0.562 | 0.721 | 0.159 | 0 | -0.562 |
WC | 8 | 5-7 | 1 | 1-3 | Gore Steals 2B | 1.211 | 1.447 | 0.236 | 0.385 | -0.826 |
WC | 8 | 6-7 | 1 | 1-3 | Strikeout Swinging Gordon Steals 2B | 0.53 | 0.626 | 0.096 | 0 | -0.53 |
WC | 9 | 6-7 | 1 | -2- | Dyson Steals 3B | 0.721 | 0.989 | 0.268 | 0 | -0.721 |
WC | 12 | 8-7 | 2 | 1-- | Colon Steals 2B | 0.245 | 0.348 | 0.103 | 0 | -0.245 |
DS-1 | 10 | 2-2 | 1 | 1-- | Gore Steals 2B | 0.562 | 0.721 | 0.159 | 0 | -0.562 |
DS-2 | 2 | 1-0 | 2 | 1-- | Gordon Steals 2B | 0.245 | 0.348 | 0.103 | 0 | -0.245 |
DS-2 | 9 | 1-1 | 2 | 1-- | Gore Steals 2B | 0.245 | 0.348 | 0.103 | 0 | -0.245 |
DS-2 | 11 | 3-1 | 2 | 1-- | Gordon Steals 2B; Gordon to 3B/Adv on E2 (throw) | 0.245 | 0.348 | 0.103 | 0 | -0.245 |
RE=run expectancy
* This value is the run Hosmer would have scored if he had successfully stolen home added to the .245 runs expected with two outs and a runner on first
In the whole, the steals would add around 1.6 expected runs (first Diff column) at the risk of losing around six runs (second Diff column) if these steal attempts weren't successful. Obviously, except in Hosmer's case, the steals have been successful. This is a reasonable expectation for a team that chose to emphasize speed, so it should be no surprise they're effective at it.
Sacrifice bunts are a different issue. As run scoring decreased, sacrifice bunts have been trending down over time because managers (and especially general managers) are acknowledging the cost of an out and not cavalierly throwing them away for just one base. The Book has been out since 2007 and the information around longer than that. The Royals have used the sac bunt five times in the postseason, all with a runner on first and no outs. The Run Expectancy Matrix suggests a runner on first with no outs would score .941 runs, whereas a runner on second with one out would score .721 runs -- the expected runs actually decrease.
Taken together, there are two sets of historical data -- teams needs to be successful in around eighty percent of steal attempts in order to justify them, and sacrifice bunts make little sense outside of very specific game situations. What the Royals have done so far is defy the odds in a major way: They've been successful in eleven of their twelve steal attempts, and two of their five postseason sac bunts translated into runs eventually scoring. This is the difference between the probability of a given event and its actual occurrence -- sometimes the odds are defied. The fact something has an eighty percent chance of failure is irrelevant in the twenty percent of occasions where it succeeds.
Entering the bottom of the eighth inning in the Wild Card game, the Royals had about a three percent chance of winning. Clearly, they defied the odds in that game, and chances are the Royals will win the ALDS -- it will be very difficult for the Angels to overcome a 0-2 deficit, especially with the next two games in Kansas City. Ned Yost is effectively playing with house money, as just about every steal and sac bunt the Royals have attempted turned out their way. Historical percentages suggest this level of success can't be maintained, but if the Royals continue to have success with these plays, the odds won't matter -- success or failure dictates the odds and not the other way around. The Royals are winning with their brand of baseball, and if that success can be sustained, they can go a long way. The odds suggest the success can't be sustained -- so don't tell them the odds.
All data from Baseball-Reference. Thanks to Tom Tango for permission to use the Run Expectancy Matrix.
Scott Lindholm is from Davenport, IA. Follow him on Twitter @ScottLindholm.