There was a moment in last night’s game when I spouted off some “traditional wisdom” about the game of baseball. I was pretty sure I was right, but it kinda caught me for a minute. I wanted to make sure I wasn’t just parroting something I heard as a kid, and never actually took the time to consider.
The moment was in the top of the 6th inning, when Nico Hoerner, having successfully stolen second base with two outs and the Cubs trailing by a run, then tried to steal third base as well. Even before the umpire signaled out, I was shouting to myself – and texting – about what a terrible decision it is to try to steal third base with two outs, unless you know FOR CERTAIN you’re going to make it. And although the play was overturned on review and Hoerner was safe, that meant it was not a “certain” situation. At least not 100%. There was apparently a very real chance of being thrown out on the attempt.
On the surface, I do get that having a runner at third base is always preferable to having a runner at second base (more ways to score). And I do understand Hoerner trying to make a little something happen with the Cubs trailing by a run and the offense struggling to really put things together. But “EVERYONE KNOWS” that you just don’t do it. You don’t try to steal third base with two outs unless it’s an absolute LOCK that you’re going to make it, since being on second base with two outs is almost as good.
Terrible, terrible, needlessly risky decision.
Er … actually … was it?
Here’s where I started to question myself, and why I wanted to dig in a little more today. We’re talking about a very good baserunner who is also a very smart player. We’re talking about a guy who plays for a very smart manager. And we’re talking about an organization that isn’t exactly lax on sabermetrically-inclined thinking.
If Hoerner thought it was a good idea, maybe Craig Counsell did, too. Maybe the Cubs have a better sense of what’s worth risking than I do, shouting old saws from my couch.
🌃So, I wanted t✅o learn. And I thought I would take you along on the ride with me.
As of typing this sentence, I have not checked the run expectancy matrices, or the pitcher/catcher battery skills at run-stopping, or any other context whatsoever. I am operating based solely on the old wisdom, and an unspecified logical approach to situational baseball. But is it wrong? Do the numbers say something we aren’t expecting? Was it actually worth taking a big risk on stealing third base, because the possible outcomes were so varied?
OK. I am going to do some looking now.
First up, the run expectancy matrices. The two I’m interested in are the one that shows you the chances a team will score at least one run in a given situation, and the one that shows you how many runs you would normally expect to score in a given situation. They both matter, and they tell you slightly different things (especially in a one-run game, where you could at least argue it was right to play that inning for just one run).
Leading sabermetrician updated run exp🌊ectancy matrices back in February, with the most recent running through the 2023 season. Here are the first two numbers of note:
That means, a successful steal like Hoerner’s increased the chances of scoring a run by 4.3 percentage points, and marked a 19.7% improvement, relatively speaking. You could argue, then, that Hoerner needed to be a bit more than 80% confident he would succeed in the steal attempt in order to justify it. That’s not precise, but it’s a way of thinking about it. Would you trade a 20% chance of being thrown out for a 20% improved chance in scoring a run? Eh. But what if it was only a 10% chance of being thrown out? Then maybe it’s a good trade.
But let’s say we’re not interested in the chances of scoring just one run. We want to know the total amount of runs you’d expect to score in each situation. Ideally, it’d be a big jump between the two situations if you’re going to risk that precious third out.
The numbers:
On a percentage bases, the improvement in run expectancy is about 13.8%. Right away, we can see where the decision to try to steal makes more sense if you’re trying ONLY to score that one run (a 19.7% bump, rather than a 13.8% bump). If you’re thinking bigger, then the risk of giving away that third out is more dramatic. How sure would you need Hoerner to be in order to justify a mere 13.8% increase in run expectancy? Well, again, you could argue that it’s something north of 86% sure. I’d have to do some more thinking and math’ing before I would say that’s correct.
Either way, the numbers are pointing in the same narrative direction: there is some wisdom in the old wisdom. Yes, stealing third base with two outs DOES improve your situation. But does it improve the situation ENOUGH to justify the risk? Only where you are at least 80 to 90% sure you’re going to succeed.
I did just look at the game particulars, by the way, thinking maybe they tipped the scales here. Like, maybe the Rays pitcher-catcher combo – plus Hoerner’s speed (83.3% stolen base rate) – made it super obvious that Hoerner WAS at least 90% likely to succeed?
Indeed, Rays catcher Ben Rorvhedt does have an abysmal caught stealing percentage – there’s a reason the Cubs ran wild last night overall! Six stolen bases! – but we know not all attempts are created equal. If you were the poor schlub behind the plate every time The Flash was on base, you’re probably going to have a really crappy caught stealing percentage.
So Statcast to create an expected caught stealing rate (based on the context) compared to what you are actually doing. The numbers here are relatively unremarkable. Rortvedt’s transfer time is well below average, but his pop time is slightly above average, and his arm strength is near the top of the league. Thus, his overall caught stealing runs compared to average is just -1, right in the meaty middle of the pack. Even on a per-throw basis, he’s once again right there in the middle of the pack. Nothing stands out.
For Rays pitcher Kevin Kelly, the numbers once again just don’t say much. the same kind of voodoo for pitchers, and Kelly rates out as average. It’s a smaller sample and pitchers are ALL really heavily clustered in the middle, though, so it’s not really a surprise that nothing stands out.
So at least by these advanced metrics, there wasn’t much that SCREAMED you absolutely must steal on these two particular guys at every opportunity. Nothing that says you shouldn’t, either, I suppose.
Where do I land now, having done the exercise? Hmm. I think maybe pretty close to where I started? Remember, I didn’t actually know how this was going to go when I started typing. I’m not making an argument.
If anything, I’m probably SLIGHTLY softer now on the whole “never steal third with two outs unless you’re certain” thing, especially if it’s a fast runner like Hoerner. I thought the numbers would be more stark than they actually were. With the right read of the pitcher and catcher, a guy like Hoerner probably CAN get himself to a 90% confidence that he’ll succeed, in which case the math does suggest it’s worth the risk. Maybe that’s exactly what last night was, and thus I would be wrong to shout from my couch.