What About LOB(%)?

It’s been a reasonable, if unspectacular, first month thus far for the Cardinals’ pitching staff. Through Tuesday’s game, they rank 9th in all of baseball in ERA and 6th in runs allowed per game (RA/9, if you will). Keeping runs off the board has helped them to a +26 run differential, 8th best in MLB. Yep, everything is juuuuust fine. Just don’t look at all of the underlying peripherals.

The underlying peripherals are where the trouble happens. Their FIP is 15th, xFIP is 17th, and K-BB% is 17th. It’s still early and there’s a lot of time for corrections, but those numbers are perfectly mediocre. Their ERA minus FIP (E-F) is -0.69, behind only Houston and Cleveland for the biggest discrepancy in the league. There would appear to be regression coming. A lot of this can be explained with their LOB% (left on base percentage). Through Tuesday, the Cardinals’ LOB% was 79.6%, fourth highest in MLB. A simple linear regression of a team’s ERA-FIP and their LOB% turns up a .651 adjusted r-squared. Here’s a scatterplot of every team since 2002, demonstrating the relationship between the two stats.

This would be wonderful news if we could expect the Cardinals to continue to strand 79.6% of baserunners. If they managed to perform that feat, they would have a higher strand rate than any team since 2002. In fact, it would a better strand rate than every other team since at least 1903, the year of the first World Series. That... is not a reasonable expectation. (Fun fact: the 2nd highest strand rate since 1903 belongs to the 2015 Cardinals, just behind the 1968 Tigers.)

More realistically, most teams have a LOB% between 70 and 72. Let’s split the difference at a LOB% of 71, and find out how many more runs the Cardinals would have allowed if they were stranding runners at an average rate. For the Cardinals to have a LOB% of 71 (more precisely, 71.3%), they would have had to have given up 94 runs. That’s 20 more runs than the 74 that they had given up through Monday’s game. Suddenly, their pythagorean record falls down to just a little above .500- 100 runs scored, 94 runs allowed resulting in 11-10 instead of their actual pythagorean record of 13-8 through Monday. Stranding runners has been worth 2 wins to them thus far.

This all begs the question- are they doing something unique that allows them to strand baserunners at a higher rate? It’s a dubious claim, at least with the information we have available. But we know the Cardinals tend to do certain things well that might enhance their strand rate. For instance, the franchise has traditionally procured very high groundball rates. Yadier Molina and the pitching staff typically control the running game better than most teams. They don’t give away a lot of walks, historically. As a franchise, they’ve done very well with preventing opponents from taking an extra base here and there. Perhaps these items can explain the high LOB% thus far.

Thankfully, Mike Podhorzer did some heavy lifting at the Hardball Times a few years ago. Using a variety of stats, he created xLOB%. By his own admission, it spits out an adjusted R-squared of .445. We’d like higher but it’s a great starting place. Let’s apply his formula to see if the Cardinals’ xLOB% tracks well with reality, or if they’re overachieving. And in case you’ve been wondering why I keep referencing 2002, it’s because Podhorzer’s formula requires pickoff data, and that’s the first year I can find it.

Unfortunately, it’s the same story. Their xLOB% thus far is 74.4%. In our 2002-2018 sample, that’s the top 89th percentile. If they could manage to keep up this pace over the course of the season, it would qualify as the 92nd best amongst all teams since 2002. That’s good! They really are exhibiting some skills that help them inflate their strand rate. But their xLOB% is still 5.25% less than their actual performance so far, making them the 2nd largest overachiever since 2002. They nestle neatly between this year’s Astros and the same 2015 Cardinals we referenced earlier, the team that drew national attention for its cluster luck and strand rates en route to 100 wins.

In real terms, if they were yielding their xLOB% so far instead of their actual LOB%, it would be 86 or 87 runs allowed instead of 74 through Monday. That’s right around one pythagorean win, nowhere near as severe as it would be if they’d simply been stranding an average rate. Strand rate regression is undoubtedly coming. But if they can continue at their current profile, it won’t be as intense as you might think.