We’re 40 games into the D-backs season, roughly a quarter of the way. The team has 37 stolen bases, ranking fifth in the majors — and to get to the sixth-place Padres, you drop all the way to 31. Even better, the D-backs have only been caught stealing 11 times so far, a 77% success rate that on the surface tells us it was worth the effort. Every player with at least three attempts has been successful at least two thirds of the time, with one exception. Last year, the team had a 72% success rate, pretty much right at the break even point — although we’re only talking about three extra successes so far this year, that’s a meaningful difference. The D-backs are running well. But are they running smart?

We’ll go a little deeper in a second on stolen bases specifically, but first, the D-backs generally have been good on the base paths in general. Baserunners have scored over 33% of the time, which ranks second in the NL — and considering the D-backs rank just sixth in on base percentage and fifth in slugging, being fast, smart or both seems to have made a pretty big difference in terms of scoring runs. Sure enough, according to Baseball Prospectus’s excellent baserunning stats, the D-backs rank second in baseball and first in the NL with 7.9 Base Running Runs. Last year, the team finished the season with just 1.8 Base Running Runs. Things can change quickly because caught stealings and TOOTBLANs move the dial down much more quickly than successes can push BRR up — but take that 7.9 BRR total for this quarter-season and drop it into last year’s results, and the D-backs would have been the best in the NL last year. By, like, a lot. This is pretty cool.

Most of that 7.9 BRR total comes from taking extra bases on outs (really, just ground ball outs) and on hits. But the D-backs are still near the top of the leaderboard in BP’s Stolen Base Runs stat with 2.01, more or less tied with the Padres for second in the NL. And considering Cincinnati’s total is almost completely because of Billy Hamilton, it looks like base stealing has been good to the D-backs.

But back to the question: are the D-backs running smart? That has little to do with how fast these guys are, and it doesn’t even depend on how good they are at stealing. A team or player that has a good feel for those things — and, basically, an expected success rate — can be smart just by adjusting accordingly.

We start with a Run Expectancy Matrix. There are 24 base-out states possible, with runners in different places and 0, 1, or 2 outs. With so many games’ worth of information, we can say that, hey, when you have the bases loaded and no outs, you’re going to score an average of 2.23 runs for the rest of that inning. If it’s a blank slate with no runners or outs, you expect to score an average of 0.46 runs. You can see the full Run Expectancy Matrix for 2014 here, at Baseball Prospectus.

We can use the Matrix to figure out what the value of a stolen base is, and that’s exactly how BP does it. Runner on first and no outs has a run expectancy of 0.82 runs. A successful steal means a runner on second and no outs (1.04 runs), and a caught stealing means one out, no one on (0.24 runs). From that, we can conclude that when there are no outs and a runner on first, a stolen base has a value of 0.22 runs, but a caught stealing carries with it a more powerful penalty of -0.58 runs.

Here’s how stolen bases and caught stealings shake out using the Run Expectancy Matrix for 2014:

No outs One out Two outs
Stealing second 0.22 -0.58 0.14 -0.39 0.10 -0.19
Stealing third 0.25 -0.80 0.27 -0.53 0.04 -0.29
Stealing home 0.17 -1.05 0.35 -0.80 0.76 -0.33
Double steal 0.47 -0.78 0.41 -0.57 0.14 -0.40

If there’s one thing to take away from this, it’s that context really matters. The double steal here refers only to the situation of first and second, with the defense going after the lead runner — but look how much damage is done if you try this with no outs and fail. The only situation in which a stolen base helps more than a caught stealing hurts is breaking for home with two outs. Stolen bases definitely help, but caught stealings are, in general, pretty terrible.

What we need from this, though, is to figure out that break even point. For each of the above situations, here are the success rates necessary to make it a good idea to run:

Event No outs One out Two outs
Stealing second 66% 74% 66%
Stealing third 76% 66% 88%
Stealing home 86% 70% 30%
Double steal 62% 58% 74%

These break even rates are for any runner, but they aren’t sensitive to the score (sometimes scoring a single run is the only thing that matters), and they’re true if you’re expecting Joe Average in the next few spots in the batting order. We’d have to think of a “true” break even rate taking account of the latter thing — for any runner, the break even rate for attempting to steal home on two outs is a lot lower if the batter is Nick Ahmed or a pitcher than it is if it’s Paul Goldschmidt.

Being “smart,” then, is not about slavishly following these numbers. But it does help to have them in mind. The questions should be: what is Player X’s chance of taking this base if he goes for it? Is that higher or lower than the “true” break even rate, which is one of the above rates but with an adjustment for the batters due up?

So, context is important. I rolled through the D-backs game logs for each game so far this season, to see if anything stuck out. The results are in the table below, at the bottom of this post. Note that I clearly missed some along the way, but I’m not casting an overall verdict anyway (the BP stats do that for us). What I didn’t do is find circumstances when the team didn’t attempt a steal but should have — but just in terms of when the attempts were actually made, I don’t see anything alarming at all. Totally smart. You can go look at the table and see what you think (let me know!), but here are a few of my thoughts:

1. Goldy only ever tries for second, which we more or less knew. Since he hits third, though, it looks like he’s found himself on first base with two outs pretty often, and except for a steal with no outs back on April 7, he’s always done so with two outs. So we shouldn’t compare Goldy’s steal percentage to the overall break even rate of about 70% — we should compare it to the 66% rate from stealing second with two outs in the table above. In Goldy’s near-MVP 2013, he had 7 caught stealings to 15 steals, and that looked not too great — it was a 68% success rate. The difference in the break even percentage used turns a negative into a positive. Cool, right? And it doesn’t hurt to know that while Goldy steals more bases than the average first baseman, he goes for it in a situation where it makes a ton of sense.

2. The D-backs may be attempting to steal second with one out a little too often. The table has 8 instances of that. There’s only one caught stealing and it is always about the other context, but 74% is a pretty high break even rate, and the team might consider tamping down players’ enthusiasm for trying to steal in that situation.

3. The D-backs have attempted to steal third 8 times, and have been caught twice. That 75% success rate is not bad. None were with 2 outs, which is good, because according to the break even table, that is pretty much the dumbest time to try. Even one out is a high threshold. But that’s the good news! 7 of those 8 attempts to steal third have been on one out, when the break even rate is 66%. That’s smart baseball, friends. I like it.

4. Stealing home is context driven, with no hard and fast rule other than “don’t do it with no outs.” I think that 30% break even rate with two outs is remarkable, and probably something for every player capable of taking a base to keep in mind when standing out there. But other than that, I think we can’t really put steals of home under a microscope from a decision-making perspective. It’s always context. A.J. Pollock‘s steal of home two days ago was a delayed steal, and doesn’t really fit into this analysis — but for what it’s worth, it did come with one out.

5. Where are the double steals? I saw two that could have been considered double steals, but while both were in a way successful, both involved errors. According to the break even table, double steals are a pretty damned good idea, especially with exactly one out. I think our natural feelings in a first-and-second, one-out situation is to hold tight and hope for the best, but the worst can include double plays, and a single in the next at bat probably scores exactly one run whether you try and fail or don’t try. A successful double steal, on the other hand, gives you the potential to score two runs on a single, and it also gives you that one amazing thing — a runner on third with fewer than two outs. Say Ender Inciarte and Pollock reach, and Goldy strikes out. So long as all of the other context isn’t going against the team (time to the plate, the catcher, etc.), a double steal seems like a really good idea. This first-and-second, one-out situation has already come up for the D-backs 37 times this year, and as far as I can tell, they’ve only tried the double steal twice.

Nothing is worse than watching a close game and seeing a bad base running decision go badly, but the D-backs have done really well to avoid that this season while simultaneously staying pretty aggressive. Here’s hoping that continues.

Player Date Event Outs SB Due Up
Pacheco 5/21 Second 1 SB Pennington
Hill 5/20 Second 0 SB Gosewisch
Ahmed 5/20 Third 1 SB Pitcher
Peralta 5/20 Second 1 SB Tomas
Pollock 5/20 Home 1 SB Tomas
Gosewisch 5/20 Second 1 SB Pitcher
Pollock 5/20 Second 0 SB Goldschmidt
Pollock 5/20 Third 1 SB Peralta
Inciarte 5/19 Second 2 CS Peralta
Owings 5/18 Second 1 SB Gosewisch
Inciarte 5/17 Second 1 SB Trumbo
Inciarte 5/13 Second 0 CS Trumbo
Goldschmidt 5/9 Second 2 CS Peralta
Tomas 5/8 Second 0 SB Hill
Pollock 5/8 Second 2 CS Peralta
Goldschmidt 5/6 Second 2 SB Pollock
Inciarte 5/6 Second 2 SB Trumbo
Pollock 5/6 Second 0 SB Tomas
Gosewisch 5/3 Second 2 CS Ahmed
Inciarte 4/29 Third 1 SB Trumbo
Goldschmidt 4/27 Second 2 SB Trumbo
Goldschmidt 4/26 Second 2 SB Trumbo
Owings 4/25 Second 2 SB Gosewisch
Ahmed 4/25 Second 1 CS Pennington
Tomas 4/24 Second 2 SB Owings
Pollock 4/22 Second 1 SB Trumbo
Owings 4/19 Second 2 SB Ahmed
Owings 4/17 Second 0 SB Ahmed
Pollock 4/17 Second 2 SB Goldschmidt
Inciarte 4/16 Second 1 SB Goldschmidt
Inciarte 4/15 Third 1 SB Goldschmidt
Inciarte 4/15 Third 1 CS Goldschmidt
Pollock 4/15 Second 2 CS Trumbo
Goldschmidt 4/13 Second 2 SB Peralta
Lamb 4/13 Second 1 SB Hill
Lamb 4/12 Second 2 CS Gosewisch
Bradley 4/11 Second 2 CS Pollock
Pollock 4/11 Third 0 CS Goldschmidt
Pennington 4/10 Third 1 SB Inciarte
Goldschmidt 4/7 Second 0 SB Peralta
Tagged with:

4 Responses to D-backs Aggressive With Stolen Bases, Situationally Smart as Well

  1. Joseph Gonzol says:

    I’m just curious, are you saying that Dbacks should steal second with one out less often because the threshold is unsustainable? 7 out of 8 is 87.5% but definitely a small sample

    • Ryan P. Morrison says:

      I don’t know if the success rate is sustainable, but it’s more that it’s just not worth the risk. Given those break even rates, you’d want to see the team try to steal second much more often with no or two outs than with one out, but with the D-backs it’s pretty even across the board, especially if you subtract Goldy’s habits.

      In other words, it was the 8 I was taking issue with, not where the 7 falls. They have an overall success rate of 77% or so, and that might be a little more predictive of what happens next than the 87.5% on the 8. Either way, just way too close to the break even point. You could say that the D-backs were just being more selective with one-out 2nd base steal attempts, but I think probably every team is selective.

      These guys are successful enough in general that maybe they just green light almost everyone, but only for second base if there are no or two outs, and only for third base if there’s exactly one.

      • Joseph Gonzol says:

        Okay, yea I think being just barely above break-even is not where you want to be if you have a high number of attempts. It’s a good point that the nature of having a higher percentage means there is less to gain.

        The most interesting number was the double steals with one out. I never really thought about that! Thanks

  2. […] to calculate some baserunning break even rates a bit ago, and that has more of an explanation of how run expectancy works. Suffice it to say, we know based on enormous, recent samples how many runs score on average in […]

Leave a Reply

Your email address will not be published.