The 2016 bullpen struggles: it's déjà vu all over again

Any casual observer of the 2016 Blue Jays season will be aware of their late inning struggles one month into the season. Countless times, they've let winnable games slip through their fingers, and while that's bound to happen, it`s been frustratingly common.

To quantify the extent of this damage, for each of the 26 games played thus far, I pulled the Jays' win expectancy (WE, also called win probability or WP) after the 5th, 6th, 7th, and 8th innings. By taking the average of each of these innings, and comparing to both the starting win expectancy (by definition, 50%) and actual winning percentage, we can drill down into where things are going wrong:

Over the first five innings, the Jays have been extremely successful, with an expected winning percentage of 0.605, which works out to 98 wins over a full season. Over 26 games, it works out to 15.7 expected wins, which compared to the actual total of 12 wins means that almost 4 wins have gone missing between the end of the 5th inning and the end of the game. I'm going to go way out on a limb and say it would be nice to have those in the bank. Incidentally, it would also put the Jays first in the AL East, before considering that most of the wins would be at the expense of division rivals.

Each of the subsequent innings has been a net loser for the Jays, with the 7th inning and the 9th onwards being particularly awful, each period costing around 5% per game and more than one expected win. The obvious culprit is the bullpen, many of whose members have struggled or at least had some ill-timed blow-ups, though the offensive side should not be absolved of blame as there have been times the bats have gone silent after the earlygoing.

The bullpen's been okay at keeping runs off the board

What's interesting is that overall, the bullpen hasn't actually pitched that badly. As a whole, they have an ERA of 4.04, which only ranks 12th worst of the 30 MLB teams and lags behind metrics such as FIP (3.78, 15th best) and xFIP (3.68, 11th best). So the first bit of good news is, we should expect the bullpen to be better at preventing runs just based on what they've done, nevermind if and when some of the established relievers turn around their season.

But despite being pretty middling in terms of preventing runs, both in terms of process and results, the end result in terms of wins and losses has been brutal. The bullpen ranks 29th in baseball in terms of Win Probability Added, with -1.43 WPA (or perhaps more aptly, 1.43 wins subtracted). By contrast, the 28th (Braves) and 30th (Reds) teams have the worst and 3rd worst team bullpen ERA. With the bullpen ERA the Jays have posted, we shouldn't be expecting miracles, but it shouldn't be nearly this bad.

We've seen this before...

Unfortunately, this is not new at all. At the All-Star break last year, I wrote about how the Jays had an historically awful relationship between their bullpen's adjusted ERA (ERA-, adjusted for park and league) and WPA. Basically, as should be intuitive, at the team-season level there's a very strong relationship between preventing runs and converting that into wins, with over half of the variation in WPA explained by adjusted ERA. Based on their first half ERA, the Jays would be expected to have about +1.6 WPA; instead they were at -3, which is a 4-to-5 win swing over 91 games. As I noted, it would have put them near the top of the division at the All-Star break. And extrapolating that pace over a full season, the Jays would have projected to underperform by about 8 wins, which would have been the 2nd worst of the 390 team-seasons from 2002-14.

Given the incredible run the Jays had in the second half, one might be inclined to think things went a lot better, especially after the Jays overhauled the pen by adding Aaron Sanchez, LaTroy Hawkins, and Mark Lowe. Alas, not really. The bullpen was even better, with a 82 ERA- compared to 89 in the first half, and posted a positive WPA at +0.51. But that was once again a significant underperformance, since the expectation based on their adjusted ERA was 1.94 WPA - roughly another 1.5 wins disappeared.

For 2015 as whole, the bullpen's WPA underperformed their ERA by 6 wins, which would slot in at 7th worst on that list of 390 team-seasons, basically at the 2nd percentile or about two standard deviations below average. And now here we are in 2016, once again underperforming. Since the beginning of 2015, the Blue Jay bullpen has the 12 best adjusted ERA in baseball at 88. They have the 3rd worst WPA at -3.92. The expected WPA over 188 games would be around +3.50: that's a swing of 7.5 wins, roughly one per month or every 25 games. That's potentially the difference between winning the division and not even making the play-in game.

I don't really have a good explanation for what's going on. Some of it is undoubtedly random stuff - a confluence of ill-timed blowups in important situations, lack of "clutch", etc. But to some extent, I wonder how John Gibbons figured into this. Managers are always second-guessed, and few if any MLB managers will manage their bullpens in a way new thinking would consider optimal. But he does have some clear faults: managing pretty rigidly to the save, underusing some relievers and misusing others, as well as cycling through multiple relievers when it may not be necessary. But that's really a topic for another day.