Statistical analysts have been saying for years now that stolen bases are not only overvalued, but that caught stealing is a tremendous loss of value. The logic behind that statement seems pretty solid. A team only gets 27 outs to score runs, and getting caught stealing takes one of those precious few outs away. Additionally, stolen bases don’t immediately translate into runs. As a very basic marker, take a look at two Mets’ seasons.

The team led baseball in 2007 by stealing an even 200 bags, thanks largely to **Jose Reyes**’ 78. The Mets scored 804 runs that year, 10^{th} best in baseball. However, in 2009 the Mets struggled offensively with Reyes hurt and scored only 671 runs for 25^{th} in baseball. The team had 122 SBs though, still sixth best that year. While the total stolen bases are obviously quite different, the success rate is the more important stat to consider. In 2007, the Mets stole 78 more bases than they did in 2009, and were only caught two more times. That SB/CS percentage is key element to consider.

Tom Tango, Mitchel Lichtman and Andrew Dolphin did some outstanding research in *The Book: Playing the Percentages in Baseball* and determined that the average break-even point for stolen base success is just above 73%. However, that rate changes based both on the game situation and the overall run-scoring environment.

Going by that 73% rate, the Mets attempted 246 steals in 2007 for an 81% stolen base rate, which is fantastic in any run-scoring environment in MLB history. Meanwhile in 2009, the Mets attempted 166 steals and were successful 74%. That season, their stolen base success rate was barely above the break-even point suggested by “*The Book*.” That year the Mets scored fewer runs, and it’s hard to suggest that their stolen bases played a pivotal role in the runs they did score.

Since the percentages don’t show if those steals turned into runs, let’s instead try a game situation breakdown to hopefully shed more light on the particulars of the stolen base.

We’ll take Reyes’ 13 multi-steal games from 2007 to do a very small game situation analysis. In those 13 games, he stole 28 total bases, and scored 11 times. Even if all 11 runs were a direct result of a stolen base, that’s just a 39% rate of success. Expand that to include all 43 multi-steal games Reyes had during his dominant years of 2005-2008, and he stole a combined 92 bases while scoring an assumed maximum of 46 runs from those SBs. That’s at best a 50% rate of a SB leading to a run. While Reyes had an outstanding 79% SB rate during that four year period, the likelihood that those stolen bases led directly to runs was quite a bit lower.

A better way of examining this is by looking at the Win Probability Added (WPA) which according to FanGraphs “is the difference in Win Expectancy (WE) between the start of the play and the end of the play.” It is important because it evaluates each situation in a game, and over the course of a season will add up to produce a total number for each player. Take **Eric Young**, Jr.’s 2013 stats for instance. He had 46 steals and was caught stealing 11 times for an 81% stolen base rate. Add up the WPA for each attempt, and you get a .475 WPA, or almost half a win. By contrast, Jose Reyes had a 1.519 WPA in 2007, his best year for stolen bases.

So we learned that the break-even point for stolen bases is above 73%. We’ve also concluded that even if a player exceeds that point, his likelihood of scoring as a result of a stolen base is far lower. So why do teams still steal? Because in a close game with the right guy on base, that stolen base *can* lead to a run and *can* lead to a victory. If you have a guy like Reyes or Young, Jr. who stands a roughly 80% chance of stealing, it makes sense that a manager would try the steal. But letting a player run whenever he wants seems to be going extinct.

WPA is a descriptive stat, not a predictive stat, meaning it tells you what happened and not what’s likely to happen in the future. Young Jr.’s WPA on stolen base attempts in 2013 tells us nothing about what his WPA might be in 2014. However, given his history and current age, it does not seem a stretch to assume that if Young, Jr. can get another 600 PA that he could reproduce his 2013 numbers. Getting those 600 PA is the big question mark though. Unless there is an injury or another outfielder falters, Young, Jr.’s playing time will be drastically reduced in 2014, taking away his chance to lead the league in steals again.

At the moment it looks like he will be spot starting in the outfield and used as a pinch-hitter. And he’ll be the first choice in traditional pinch-running situations. It remains to be seen if the **Terry Collins** will be more aggressive in utilizing a pinch-runner with a weapon like Young Jr. available on the bench. It might be the best way to optimize his main skill (speed) while minimizing the impact of his low OBP and below-average defense.

It turns out stolen bases are not simply overvalued. They can be tremendously meaningful when used judicially, like in tight scoring games where that one run can mean the difference between a win and a loss. It means Collins using Young, Jr. for what he’s made to do, steal bases. Otherwise you’re either wasting outs when you don’t have to, or just padding someone’s stats, and neither leads to a championship.

I have to say that I’m shocked that Tango has the average break-even for SB at over 73% – that number feels too high to me. I have no doubt that it’s carefully researched and I’m not qualified to argue with him over that. Just saying for my own personal POV — that number seems high for today’s run environment.

What really interests me here are the WPA numbers. It’s all well and good to have average numbers but it’s very nice to have WPA so you can see the context that the steals were attempted in. As for EY – his SB/CS numbers are very close – but under – what the average numbers tell us. Using the runCS average values from FG, Young’s 46/11 mark translates to 4.976 runs or just under half a win. His WPA in those events came out to .475

WPA is measured in Wins, not runs. So, a WPA of 1.0 would mean 1 Win or 10 runs.

Meanwhile, 2007 Reyes was the opposite of 2013 EY. His actual WPA was higher, much higher, than the average SB/CS numbers would indicate. Patrick has his WPA on stolen base attempts at 1.519 while the average numbers indicate 6.507 runs. So the net worth of Reyes’ steals-caught stealing that year were over twice as valuable as what the average suggests. That’s incredible.

I’m new to this stat, but wouldn’t the WPA be higher for a stolen base if it was swiped with 0 outs as opposed to 2 outs. If that was the case, stealing a timely bag with 2 outs might be devalued somewhat because of the probability of scoring thereafter. I wonder if that’s why Reyes’ is so high. I mean he always had the green light regardless…a few rare instances aside.

Good stuff here. SBs are an asset many teams can’t claim. Reyes was that special player who lived on 2nd and 3rd base because of his speed to steal bases, and high extra base hit totals. An rare and elite weapon.

Stating the obvious here, but EY would’ve scored more runs in 2007, and Reyes less in 2013 because of the strengh of lineups. I mean I can only imagine these stats applied to the 1999 team with Henderson and Cedeno combining for over 100 SBs in one of their best lineups ever. I’m not sure penalizing a guy for not scoring is fair if he succeeds at stealing.

S

Then you should pay extra attention to the WPA numbers, which don’t reward/penalize a guy for scoring/not scoring. It just looks at the Win Expectancy of the stolen base/caught stealing.

I actually did do that. Interesting stuff. I hope EY can get on base more, maybe that number will grow