Anybody Got a ‘Pen?

A little less than two weeks ago, Travis Sawchik and Craig Edwards wrapped up our positional power rankings series by taking a look at each team’s bullpen as composed at the start of the season. In Craig’s introduction, he noted that this was, in one sense, a bit of a fruitless exercise. Bullpen performance is very poorly correlated year to year. A combination of midseason acquisitions, injuries, and just plain bad luck can have an outsize impact on end-of-year results.

But the unpredictability of a bullpen’s performance in the future is another matter altogether from the performance of that bullpen in the past. Relievers threw a little over 38% of all innings pitched last year, and that figure is up to 42.3% through games played this past Saturday. Having a good — or at least not a terribly bad — bullpen is increasingly critical to a team’s chances of making and thereafter succeeding in the postseason, and so even if we should retain a measure of humility about our ability to predict what will happen before the season, we should nonetheless keep a close eye on how bullpens actually do once the season starts.

So let’s start looking. Here’s a (sortable) table for your consideration:

What this table shows you is each team’s projected relief K/9 and BB/9, as of the time we published our positional power rankings for relievers; each team’s actual relief K/9 and BB/9, through games played on April 7th; the difference between those figures in raw terms; and that difference divided by the initial projection, so that we can easily compare each relief corp’s variance to date from our expectations for them before the season. In this case, I’ve presented an increased walk rate as negative variance, because it’s worse for the team. So, for strikeouts and walks, positive is good, negative bad.

You may find it easier to take in this information in the form of a pretty picture, so I’ve made a pretty picture for you, too. Here, the last two columns of the table above are plotted on the X and Y axes, respectively, so that teams to the right of the center line have improved upon their K/9 projections so far, and teams to the left have declined from their projections in that statistic. Teams above the horizontal line, meanwhile, have improved upon their BB/9 projections to date, whilst teams below that line have walked more than we might have expected coming into the season.

I like this chart because it strips out some of the sequencing issues that have presumably driven some of the early WAR figures for bullpens. Cub relievers have stranded 90% of runners on their watch, for example, which probably won’t last all season, and which is nonetheless driving some of their bullpen’s league-leading 1.02 ERA and 1.0 WAR. Here, we see that the Cubs are performing a little bit better than we expected — striking out about 10 batters every nine innings and walking about four, which is very good overall — but perhaps not quite as much better than we expected as just looking at ERA or WAR would have you believe.

I also like it because it allows us to easily identify the teams whose relievers are both striking out more batters and walking fewer than we’d expected them to (the teams in gold) and those who are striking out fewer and walking more (the teams in red). Most teams, of course, are doing some things well and some things poorly. But up in the top right corner are the Mariners, and it’s hard to deny that it’s been kind of fun to watch Edwin Díaz and James Pazos strike out 12 of the 23 batters they’ve faced for the Mariners in the early going, especially when we expected both men to be doing only about two-thirds as well as that when the season began.

Now, I’ve got to be careful. My old Baseball Prospectus colleague Russell Carleton recently cautioned against reading too much into his famous “stabilize by” charts, which for many years allowed writers like myself to make far-too-bold claims about early-season performance based on a remarkably small number of plate appearances or batters faced. As a result, you won’t find me out here reading too much into what we’ve seen in the early going so far. Each relief corps has only faced three big-league teams at this point, and one team’s squad (the Royals’) has not yet faced its 80th batter of the season. (All other teams have faced at least 99, and the Marlins are nearing 200.) The next 150 batters each team’s relievers face will look different than the last 150, and that will matter for these numbers going forward.

So I’m not here to say that these numbers are especially predictive of the future. They’re not. And they’re certainly not for individual relievers like Díaz and Pazos. Drop three of the 12 Ks those two men have gotten so far and the numbers start looking really different — and all it takes to make that kind of difference is one tiny adjustment, or one or two hitters who ate a particularly dodgy chicken salad the day before the game. So this isn’t a “the Mariners are going to have the best bullpen in baseball in 2018” post. This is a “the Mariners have, so far, struck out many more hitters than we’d thought they would, and walked fewer” post. What does that kind of post mean? How should it be read?

Well, that kind of post can and should guide us to look a little closer at the teams and players that are surprising us, and try to see if there’s anything in what they’re doing that’s different than what we expected of them. If not, there’s probably little reason to expect any anomalous performance we’ve observed to continue. If so, then although we can’t say for sure that the surprising results we’ve seen are due to the changed approach, we can keep watching, and see what transpires down the road. So, without any expectation of future performance anything like performance in the past 12 days, I’ll note here that Edwin Díaz has changed his vertical release point this year, after staying basically constant every year of his career to date. James Pazos has not, as far as I can tell, been doing anything especially different at all this year. Let’s keep watching together, shall we?