Sometimes you think big. You have hypotheses or theories about how the game of baseball works in some fundamental way or you have a deep analysis of a player or a team or transaction that shines a light nobody has yet shined.

Other times, you ask resident data guru Tim Collins for some help and he delivers and you just want to explore the results in public.

Guess which type of article this is.

The data I asked for and received takes our Team Baserunning report and flips it: How many baserunning runs did teams allow or prevent on defense in 2012? I have the whole answer, but for now, I want to explore one tiny corner of it: Here are the two teams that allowed the most runs above (i.e. worse than) average on stolen bases.

The Pirates probably make you shrug, but the Twins are a surprise. Joe Mauer! Joe Mauer's magnificent scalp! The Minnesota golden boy, though, had a terrible year trying to stop the opposition's larceny, throwing out a measly 14 percent of basestealers. Moreover, Mauer donned armor for less than half of his team's innings, with the rest being handled by Ryan Doumit and Drew Butera, who themselves combined to catch only 22 percent of thieves.1

Catchers, of course, make up only a portion of the battery tasked with stopping base-stealers and, per Ben Goessling last year, Ron Gardenhire laid the blame squarely at the feet of the moundsmen, noting that many of them were slow to the plate. Avoiding the slide-step or otherwise rushing the delivery can be a valid trade if a hurler gets better pitches and thus better pitching performance out of it, but:

Pitcher SBA CS% FIP
Alex Burnett 9 0% 3.98
Sam Deduno 9 11% 5.45
Carl Pavano 8 13% 4.33
Francisco Liriano 15 13% 4.20
PJ Walters 7 14% 5.43
Jeff Gray 6 17% 5.80
Nick Blackburn 6 17% 6.05
Brian Duensing 6 17% 3.77
Scott Diamond 19 21% 3.89

In an American League that scored 4.45 runs per game, those aren't great figures for a bunch of guys who may, if you believe Gardenhire, be taking their sweet time getting the ball into the mitt.

The pitcher's body isn't the only thing moving, though. What about the Twins' notorious (though changing) predilection for soft-tossers?

Pitcher Fastball %2 Fastball Velocity
Alex Burnett 63% 93-94
Sam Deduno 53% 91
Carl Pavano 61% 87
Francisco Liriano 50% 94
PJ Walters 62% 89
Jeff Gray 55% 93-94
Nick Blackburn 69% 90
Brian Duensing 58% 92
Scott Diamond 60% 90

I'm aware of some work examining stolen bases using PITCHf/x data, but not as much as you'd think, so I can't tell you a lot at the moment about whether we would expect better or worse basestealing numbers given these particular pitchers' arsenals and average velocity. But I can tell you that major-league pitchers threw about 62 percent hard stuff last year and averaged around 91 mph with it.

For a different kind of context, a Carl Pavano pitch at 87 mph is traveling 127.6 feet per second. Pitchers generally release the ball about 55 feet from the plate. Thus, we're talking about 0.43 seconds from release to plate. Add 10 mph to that fastball and you shave off about 0.04 seconds. That's not nothing when you've got Ron Gardenhire aiming for 1.2 second deliveries and complaining that the pitchers are clocking in at 1.4 instead, but it's a tiny portion of the overall issue. Now, a 70 mph curve instead of a 92 mph fastball gets you well up over a tenth of a second (and presumably introduces more difficulties for the catcher in fielding the ball cleanly to boot), so the pitchers like Deduno and Gray might actually be suffering (and causing their poor catchers to suffer) from their pitch selection even though both have perfectly good velocity. But overall, the average velocity and a rough cut at pitch selection for the Twins staff doesn't jump out as obviously causing stolen-base difficulties.

Handedness might also matter. The Twins had six left-handed pitchers appear for them in 2012 and lefties represented 33 percent of the team's total batters faced. "Well, that's low!" I said to myself, having hit upon an explanation for why Twins pitchers are apparently so poor at holding runners. But there are my exclamations and then there is data, which shows that 31 percent of plate appearances in the American League last year occurred with a lefty on the mound. And also that three of the Twins' lefties appear in the table above, and not just any three but the three (Liriano, Duensing, Diamond) who faced the most batters of all the lefties on the team. So okay, there goes the handedness idea.

I'm feeling pretty comfortable at this point following Ron Gardenhire into the foxhole of "blame the pitchers for slow deliveries." We know Joe Mauer can throw and we've been told that Drew Butera can as well. The pitchers aren't any less left-handed than other staffs, they don't really throw fewer fastballs, and they don't throw substantially slower.

Now, while acknowledging that a more granular approach to the PITCHf/x data might yield more insights than we've got above (but might also yield a book), let's flip to a different aspect of the Twins' theft-permitting proclivities.

The Twins were not actually last in baseball in caught stealing percentage, as Washington and Pittsburgh finished behind Minnesota. And while they did not have the worst pitching in the league in terms of putting runners on base, they were fifth in stolen base opportunities faced, with 93 more than the league average.3 Put a third-worst percentage together with a fifth-worst opportunities faced and the Twins should be right up there in total stolen bases allowed, right? First or second or third or fourth or fifth or sixth? Or seventh? No. Make it 14th, just three more than the league average.

This tells me in a very general way that, despite putting more runs on the board via the steal than against any other team, the Twins' opponents might not have stolen enough. Or maybe what it tells me is that the Twins were getting their collective ass whomped so bad last year that nobody needed to bother stealing. Minnesota allowed 131 more runs than they scored. They won 66 games. They lost by at least five runs 29 times. So that's one possibility, that if they'd somehow managed to be a good team while maintaining this skill at stopping the stolen base, they'd have allowed even more.

Another possibility could be the lag of scouting reports. How long did it take for managers to be convinced that Twins pitchers were slow enough that they should take advantage with more steals even when Joe Mauer was behind the plate?

Month SBA/PA
April/March .023
May .026
June .018
July .013
August .028
September/October .022

Never, I guess. Okay. Look, I tried. What do you want from me? Let's just blame the blowouts after all and see the Twins' total stolen bases allowed as the result of unwritten mercy rules.

I want to mention one other issue here because it came up while I was researching this and it made me alarmed and I don't want you to be alarmed. What made me alarmed is this:

Team SB Runs SB CS
Minnesota 7.2 111 24
Pittsburgh 5.8 154 19

You can see why I'm alarmed. The pitiful Pirates backstops/batteries (Rod Barajas and Michael McKenry on the one hand; A.J. Burnett and Jared Hughes on the other) gave up 33 more steals and killed five fewer baserunners (despite being pirates), yet our baserunning reports say that Minnesota's stolen bases allowed were worth more runs to the opposition. Maybe you wouldn't have noticed this had I not mentioned it, but just in case you did, here's a little math to show that this is in fact a perfectly possible outcome, as it all comes down to when the steals occur.

The basic method of our baserunning reports is linear weights. This is the 2012 run expectancy table that shows the average number of runs scored from all the possible base-out states. Permit me to make a simplified summary of what stealing and getting caught is worth:

Outs SB24 SB3 CS2 CS3
0 +0.22 +0.23 -0.59 -0.81
1 +0.14 +0.24 -0.41 -0.55
2 +0.09 +0.05 -0.22 -0.315

Using the above numbers, we can look at the minimum and maximum for the two teams in question. That is, what if every steal of second and every steal of third that we know each team allowed were worth the most possible runs and every caught stealing were worth the least possible runs? And what if vice versa? This is what:

Team Maximum Minimum
Minnesota 19.256 -4.79
Pittsburgh 30.02 1.21

I googled "timing is everything" hoping it was a quote from Shakespeare or David Mamet or something but I guess it's just a shopworn cliche without a pedigree worth mentioning. That doesn't make it less true, though, and I hope that the above table provides a clear illustration of the idea. Stealing and getting caught at the best and worst possible times can swing the value of that baserunning by upward of 25 runs over the course of the season.

Which means we can answer our question. How could the Pirates, with their horrendous rates of caught stealing and their significantly higher volume of steals allowed, actually have lost fewer runs to their opponents in stealing?


  1. There was also Chris Herrmann, who caught 19 innings and allowed one steal in one attempt. That doesn't help the cause, so it merits mentioning, but, as you can see, it does not merit more than a footnote. â†©

  2. By this I mean four-seam, sinkers, and cutters. All the stuff that's relatively hard and relatively straight. â†©

  3. This comes from Baseball-Reference and is defined simply: Was there a runner on first or second with the next base open? â†©

  4. "SB2" means a steal of second. The other symbols in the table should hopefully follow naturally. I'm also, more importantly, defining this in a stripped-down way: the only runner on base is the one who is stealing. That is, "SB2" really means "going from the base-state 100 (runner on first) to 020 (runner on second)." â†©

  5. If you needed another illustration of why break-even rates are what they are, by the way, here you go. The cost of getting caught far exceeds the gain of stealing the base. â†©

  6. I'll provide this one example and you can see how it applies to the other three fields in the table:

    101 steals of second times 0.22, the highest run value for a steal of second

    10 steals of third times 0.24, the highest run value for a steal of third

    23 caught stealing second times -0.22, the least bad run value for a caught stealing of second

    1 caught stealing third times -0.31, the least bad run value for a caught stealing of third

    This gets you 22.22 + 2.4 – 5.06 – 0.31 = 19.25. That is, the Twins could have allowed 19.25 runs worse than average on base-stealing had all the steals and the caught-stealings happened at exactly the wrong (from the Twins' perspective) time. â†©