Matheny's bullpen usage by leverage in 2016

It might be hard to remember now, but in the first half of the season, the bullpen was kind of a mess. Tyler Lyons was giving up homers all over the place, Trevor Rosenthal was walking everyone in sight, and Seth Maness appeared (and indeed turned out to be) broken. Seung Hwan Oh was the lone brightspot, but it took a while for him to gain the closer's role amid some mighty struggles from Rosenthal and the other late-inning options.

Bullpen usage is one of the most common critiques a fan will make of a manager. Using your best relievers in the most important situations is key. Every situation has a certain amount of leverage involved. That leverage is quantified based on the inning, score, outs, and base runners at the time. A 2-1 game in the eighth with one out and runners on second and third will have much higher leverage than a 7-3 game in the fifth with no one on and two outs. If you're interested, here's a table hosted online that has each situation and the associated leverage.

For this analysis, I decided to see how well two stats matched up for each reliever: xFIP and gmLi. FIP is the go-to stat typically for pitchers, which is based on strikeouts, walks, and homers. xFIP on the other hand deals with strikeouts, walks, and fly-balls. They are calculated the same except a league average HR/FB rate is used to estimate the amount of homers a pitcher will give up. Because HR/FB rate takes longer than a single reliever season to stabilize, I prefer to use xFIP in these situations.

gmLi stands for Game Leverage index, and an explanation is linked above when I first mentioned leverage. It represents the average leverage a reliever has when entering the game. Thanks to fangraphs.com's leaderboards,  Here is every Cardinals pitcher that threw at least ten innings out of the bullpen in 2016, along with their xFIP and gmLi:

The first line is the worst part: Kevin Siegrist entered into more high leverage situations than any other Cardinal reliever, and had the third worse xFIP. Luckily, that didn't cost the Cardinals, as Siegrist had good results, to the tune of a 2.77 ERA. That low ERA likely explains why Siegrist had the high leverage in the first place. Perhaps Siegrist has some ability to beat his FIP and xFIP, but certainly not to the degree he did in 2016.

It's nice to see that Jonathon Broxton's leverage was so low. It's a common complaint around VEB that Broxton is used in too many important situations, and I had the same opinion before looking at the numbers myself. His first and second half splits are pretty close too, so it's not like Matheny finally learned how to use Broxton in the second half. So props to Matheny on that.

Perhaps this chart would be clearer though if we presented it in a slightly different way. I ranked the Cardinals' relievers by their respective xFIP and gmLi. Then, I took their xFIP rank minus their gmLi rank to find the difference. So, if a pitcher had the third best xFIP but the fifth highest gmLi, their difference score would be -2. Here are the results:

We see that Siegrist's usage was the most out of wack by far. By this method, Zach Duke was used in higher leverage than he should have been, but this model doesn't take into consideration that he was one of the best options against lefties. I think it's fair to critiisize Matheny for leaving Rosenthal as the closer's role too long, but these numbers indicate that Rosey's usage as a whole wasn't all that out of the line.

Seth Maness, Dean Kiekhefer, and Jerome Williams were the most correctly used relievers according to this method. Fans have worried in the past that Matheny would rely on Maness' double-play magic once he lost effectiveness, but Matheny tapered usage to the correct degree. Matheny clearly didn't have much confidence in Kiekhefer and Williams, nor should he have.

Oh was clearly the best reliever for the Cardinals, but he didn't lead the team in leverage. But what can you do? He rose to the level of bullpen-ace pretty quickly, entering the highest leverage situations before the ninth inning. As I mentioned, Mike was a bit slow to pull Rosey from the closer's role, but eventually he did, and in doing so instilled the best reliever as closer. He also started to make it a habit to bring Oh in before the ninth if needed, for 4 and 5 out saves. It's hard to critisize Matheny's usage of Oh.

Alex Reyes was the second-best reliever by xFIP, but was only used in the 5th most important set of situations. Matheny obviously values Reyes highly as he inserted him into the rotation in late September, so this seems more like the result of tactical mistakes than not valuing Reyes correctly. Or perhaps it just took Matheny some time to see Reyes' superiority.

Matt Bowman had a great season, but still struggled to work his way into high-leverage situations. At 3 spots lower on the leverage totem-pole than he should have been, and a member of the bullpen the whole season, this is probably the second biggest mistake Matheny made with regards to doling out leverage to the right pitchers. Miguel Socolovich and Tyler Lyons were both 4 spots lower than their xFIP's indicate, but both were available only for fractions of the season. Lyon's lack of use can also be attributed to a glaring homer problem, though it shouldn't be expected going forward.

Overall, I have to say Matheny doesn't grade out bad by this approach. The biggest over-use and under-use according to this method can be explained by Matheny's result-oriented approach to doling out playing time. That approach gets on my nerves, but at least he had some reason. Matheny was a little slow in replacing Rosenthal, and a little slow at utilizing Alex Reyes and Matt Bowman. Also, maybe Socolovich has a future in St. Louis that Matheny doesn't see yet. While those mistakes annoy me, hey, it could be worse.