The Oakland A’s confusing bullpen

The Oakland A’s bullpen is tied for the MLB lead in blown saves. There have been 27 games in which the pen was handed a lead after the fifth inning and then let the opponent come back to at least tie the score, and the A’s are 10-17 in those games. It’s easy to argue that this performance by the relievers is the single biggest thing holding Oakland back in the standings, and that with even average fortune in this department they could be in control of the Wild Card race instead of locked in a tight battle.

There’s one problem with that conclusion, though. Other than those blown saves, the A’s pen has great stats across the board. By almost any other measure, they’re a Top 10 unit in the majors.

What gives? Is this bullpen good or not? Heck, what even makes a bullpen good at all, the process or the results? Could there be some other factors at play that are confounding the data?

Let’s start with a quick look at the blown saves themselves. The A’s have 27, but what does that mean in context? (Actually it’s 28, but two of them came in the same game, so for overall team purposes it’s 27 games.) Remember, the A’s are good, so theoretically they should have more leads to blow than a bad team. The club they’re tied with is the Nationals, who lead the NL Wild Card with a record similar to Oakland’s.

Of the Top 10 teams in blown saves, four of them are comfortably below .500 (SEA, SD, PIT, KC), two of them are decent but unlikely to make the playoffs (BOS, NYM), and four are firmly in contention (OAK, WAS, LAD, CHC). That’s a pretty representative mix from throughout the standings, so no, the top of this leaderboard is not just reserved for winning teams who build lots of extra leads to blow. The reverse is true, too: Of the Bottom 10 in blown saves, two of them are contenders (STL, CLE) and another is close (PHI).

But that still doesn’t fully allow for quantity of opportunities. For that we need to turn to a rate, for which we’ll use blown saves divided by save/hold situations. Oakland’s 160 situations are the fourth-most, even though they’ve had only an average total workload in terms of innings, so they have indeed had more chances to blow it than most pens. There are 12 other clubs with higher rates of blown saves per situation.

Perhaps they aren’t really the worst in terms of blowing leads, it just seems that way because they’ve had so many to protect. But they’re still below average in this regard, and on top of all that, they’ve picked up another 12 losses in games where they entered with a tie and let the opponent go ahead, with no blown save involved. None of that makes sense when you consider the following numbers and their MLB rankings among the 30 teams.

• 5th in fWAR (4.9)
• 9th in ERA (4.15)
• 7th in FIP (4.15)
• 7th in xwOBA (.299)

Pick your favorite. The traditional run-prevention of ERA, the fielding-independence of FIP, the underlying batted-ball data of Statcast, or the catch-all of WAR. Oakland is great in all of them. Regarding that ERA, they’re around league average in terms of stranding inherited runners, so they aren’t secretly allowing a bunch of extra runs that get credited to the starters.

Looking slightly deeper, here are a few constituent stats:

• 13th in K% (23.8%)
• 9th-best BB% (9.3%)
• 3rd-best HR/9 (1.14)
• Best HR/FB (11.0%)

Again, strong across the board. The strikeouts are only a tick above average, but they’ve been good at limiting walks and they’ve prevented homers as well as anyone in the sport. These numbers offer no explanation for the blown saves.

How about Win Probability Added? Oakland’s pen has a value of -1.39, but the whole league is in the negative overall, so that mark is really around average. Their “Clutch” score of -3.00 is 28th (third-worst), which tells us they are performing worse in big spots than they do in an average game situation. But that’s just a confirmation of what we already knew, that they seem to be good overall but bad in the most crucial moments.

So again, what gives? How can we bridge this disconnect between the bullpen’s overall quality performance, and their weird inability to hold leads?

Here’s one big factor that could be skewing the data: Liam Hendriks. You’re surely aware he’s having a good year, based on his All-Star berth and ascension to the closer role, but you might not realize the full extent of it. He’s basically been 2018 Blake Treinen. (Note that these stats exclude Hendriks’ two games as an opener/starter.)

Treinen, 2018: 0.78 ERA, 1.82 FIP, 80⅓ ip, 100 Ks, 21 BB, 2 HR, .236 xwOBA
Hendriks, 2019: 1.54 ERA, 1.89 FIP, 70⅓ ip, 101 Ks, 17 BB, 4 HR, .231 xwOBA

Treinen had the edge in ERA, but everything else is nearly identical. In terms of fWAR, Treinen earned 3.6, and Hendriks is at 3.0 with a month left to go. Call it a wash there too. We marveled last year that Treinen had one of the best reliever seasons ever, and Hendriks is effectively repeating it.

The reason this matters is that Hendriks accounts for a huge part of the bullpen’s overall good stats. But, you might say, every team has a top reliever with good stats, and they’d look worse if you removed him! Indeed they do, but no one else has a Hendriks. (Except for the Padres, as Kirby Yates has him matched in fWAR.)

How much better are Hendriks and Yates than the rest of the league? They both stand at 3.0 fWAR, and the next-best is Aroldis Chapman at 2.1. Then come another pair at 1.9, and the difference between Hendriks/Yates and the fourth-place tie is the same as the difference between the fourth-place tie and 46th place. No pen besides the A’s (and Padres) has their numbers affected so greatly by one star arm.

And what does it look like if we do remove Hendriks from the equation?

• 1.9 fWAR (would be 17th)
• 4.60 ERA (would be 18th)
• 4.53 FIP (would be 16th)
• 21.7% Ks (would be 26th)
• 9.7% BB (would be 15th)
• 1.24 HR/9 (would be 11th)

Pretty mediocre when you cut the one huge star. It’s not so much that the A’s pen has been really good, it’s that Hendriks has been incredible. He’s carried them into the Top 10 on his own.

Again, every pen would look worse if you took away their best performer, and maybe if you did that to every team then the A’s would still eke their way toward the Top 10. But no one (except the Padres) would fall this far because of it. Over the last five seasons, there have only been seven relievers who posted at least 3.0 fWAR in a year (and only 19 who scored 2.5 or higher), so if Hendriks disappeared then his replacement would almost certainly be significantly worse. Shout out to Yusmeiro Petit, who has been darn good himself (1.2 fWAR, 2.84 ERA, 3.70 FIP), but this has truly been the Hendriks Show.

But wait. Hendriks has also blown five saves. He’s converted 18 saves and seven holds, so his 83% success rate is fine, but you’d still expect something closer to perfection from a guy having one of the better reliever seasons in recent memory. He may be carrying the team in every other stat, but even he can’t solve their lead-blowing problem.

All of this leaves us no closer to an explanation than we started. The overall bullpen probably isn’t as good as their numbers suggest, because Superstar Hendriks is skewing the data. And their MLB lead in blown saves probably isn’t as bad as it looks, because they’re really only slightly below-average in terms of conversion rate. Like it so often does, the truth probably lies somewhere in the middle of those two extremes. They might just be a mediocre pen on a team that puts together more leads than most of the rest of the league.