Intentionally Walking the Bases Loaded: A Primer

Mike Shildt had a decision to make. It was only the first inning, but the Brewers were all over Daniel Ponce de Leon. They’d already scored three times, and had runners on second and third for number eight hitter Luis Urías. A hit here could break the game open, so Shildt took a risk and intentionally walked Urías. With a pitcher batting next, maybe he could salvage the inning.

There was just one problem: Daniel Ponce de Leon was pitching. His 11.6% walk rate this year has actually lowered his career mark. Not only that, but he’d already walked a batter unintentionally this inning, though it’s unclear whether that’s predictive. Either way, though, whoops:

Maybe that was a strike, and maybe it wasn’t. In any case, it turned into a run, and the game eventually turned into a Brewers rout. The Cardinals scored only three runs; as it turns out, the first inning was all Milwaukee needed. Urías didn’t have a hit on the day, not that that’s a particularly telling statistic.

Normally, I’d break down the pros and cons of Shildt’s decision in minute detail. Avoiding Urías and his career .318 OBP to face a pitcher seems good. Turning a walk into a run with Ponce de Leon on the mound seems bad. It’s certainly not a slam dunk in either direction.

In this particular case, there’s no time to review the particulars, because Shildt had some batters to walk. On April 14, a mere three days later, he found himself in a similar bind. In the top of the second inning against the Nationals, Adam Wainwright was in an even more precarious situation. He faced second and third with no one out, with the game tied.

This time, the number seven hitter was up, but the Nats often bat their pitcher eighth, and Joe Ross lurked in the on-deck circle. Shildt again put four fingers up, juicing the bases. Wainwright managed to escape the jam — he overpowered Ross, who was attempting to bunt, then got Victor Robles to swing through a cutter.

Two free passes, two opposite results. This one was better-advised, despite the fact that the free run was more likely to matter, because Ross was far less likely than Yan Gomes, the recipient of the free pass, to take advantage of the one-out situation to score a run. Still, Shildt was living dangerously.

Again, there’s no time to break that one down, because he wasn’t done. On April 16, well, you get the idea. Second and third, one out, eight-hole hitter Mickey Moniak due up in a one-run game in the bottom of the second; Shildt knew what to do. Moniak trotted down to first, and Zach Eflin stepped to the plate. After a first pitch strike, disaster struck:

This time, the Cardinals didn’t limit it to a single run. The next four batters reached before Carlos Martínez recorded a second out, and the inning ended with a six-run Philadelphia lead. Martínez has hit a shocking eight batters this year; it’s safe to say that his command hasn’t been razor sharp. He might well have walked Moniak anyway, but choosing not to pitch to the bottom of the lineup didn’t exactly pay off here.

I could go on… and on… and on. The Cardinals have issued a whopping seven intentional walks that have loaded the bases. Facing the pitcher is an enticing incentive, but it doesn’t automatically make the decision right, and two out of those seven times were actually walks to position players.

In fact, the Cardinals are far and away the leaders when it comes to filling up the bases by choice. Here’s the leaderboard so far this year:

In all, there have been 61 walks that fit the criteria I used. They were reasonably straightforward. First, the walk had to fill the bases. Second, it had to be truly intentional; giving Bryce Harper a trip to first base after he reaches 2-0 doesn’t count. All of these walks are wholly up to the manager, a decision to skip a batter simply to face someone else.

With so many examples already this year — truly, 61 is a ton — diving deeply into each one was out of the question. Instead, I came up with a generic way of looking at each intentional walk that can give us a rough idea of who’s costing their team the most — or, hey, adding the most value, let’s not leap to any conclusions — by intervening and jamming the bases full.

First, I took the change in the win probability we report for each event in our game logs for each intentional walk. Going from second and third to the bases loaded, particularly in the early innings, adds baserunners and therefore potential runs to the opponent’s score, so most intentional walks are a bad idea. That’s not universally true, though; loading the bases with one out can increase the probability of double plays, and thus the probability of escaping the inning without any runs scoring. Late in tie games, that’s a boon.

Win probability is matchup-neutral. It doesn’t consider the identity of the pitcher, the bypassed batter, or the new batter. Many of these intentional walks bring up the pitcher, so we need to account for that. I took preseason wOBA projections for each batter, then used pitchers’ overall wOBA this year anytime the pitcher spot was due up next. That differential can be converted into context-neutral runs.

Context-neutral isn’t right here — the context is decidedly not neutral — but it’s close enough for a first pass. The next step was to convert those runs into win probability. Again, I’m approximating here, but I multiplied the runs by the average leverage index over the two plate appearances, then divided by 9.709, this year’s approximation of runs per win. This gave me a fraction of a win, which is the same unit as win probability, so I simply combined the change due to base/out state and the change due to batter identity to create the overall change in win probability.

As an example, take the intentional walk that the Blue Jays issued to Randy Arozarena on May 21. It came with two outs in the 11th inning, and moved a runner from first to second. That’s a huge negative base/out state change, because it moved a second runner into scoring position in extra innings, where a two-run lead is substantially better than a one-run lead — negative 3.1% of a win, in fact.

Brett Phillips was due up next, which changes the calculus somewhat. Arozarena’s preseason wOBA was projected 66 points above that of Phillips. That works out to 0.053 runs per plate appearance. Phillips’ at-bat came with a leverage index of 6.89. Combining those, we get an equivalent of 0.31 runs, or 3.2% of a win. This intentional walk was roughly neutral — a slight win for the Blue Jays, who issued it. Phillips struck out looking, incidentally.

In any case, it’s time to go to the numbers. Walking the bases loaded feels terrible as a fan, and it seems to always result in a walk to the next batter. Just this week, the Cardinals intentionally walked Ketel Marte for the privilege of walking Carson Kelly unintentionally, driving in a run. Here’s the thing, though. It doesn’t seem to matter all that much. Seriously, these are the smallest results imaginable:

I’ll grant that my technique could be better. Using specific probabilities of each event rather than generic wOBA could improve things, as could better accounting for the pitcher and the platoon matchup. Plenty of my frustration as a Cardinals fan (though not as an analyst) comes from the dissonance of the team with the most unintentional walks in baseball issuing more on purpose.

Overall, however, it’s apparently not so bad to intentionally walk the bases full, at least before considering the knock-on effects; more plate appearances against the good hitters in the lineup (since you’re forfeiting a chance to face one of the bad ones), more stress on pitchers, and more chance of a big inning if you can’t get the pitcher out — win probability definitely doesn’t consider which batters are due up after the one at the plate, given that it doesn’t even consider who’s at the plate.

Overall, I’m still against intentionally walking the bases loaded. Doing it 61 times to net a small gain before considering negative externalities seems like a bad plan, and intentional walks are just a bad plan overall; baserunners are more valuable than managers think they are. But I owe Mike Shildt an apology for all the times I grimaced when he did it… at least for now.