The Diamondbacks Bullpen: Good, Bad or Indifferent?

Looking at blown saves isn't really very useful, in itself, for a number of reasons, not least that teams have had a greatly-differing number of save opportunities. Only the Pirates and, again, Dodgers, have had more save chances than Arizona's 35 to date and, obviously, more save chances will tend to mean more blown saves. If you look at saves made, as a percentage of chances, Arizona looks a little better, now sitting ahead of the Dodgers, Mets and Cubs at 60%. With the league rate at 66%, that means we have about 3.6 more blown saves than an "average" team, which is still not good.

But the weirdness of the save statistic continues. That percentage is based only off blown saves and successful saves, what are known as "save opportunities". You can only get a save if you are the last pitcher used, but blown saves can happen to any reliever - indeed, the majority for Arizona this year, have come before the ninth, happening as early as the sixth inning. Those are "save situations": Except, if a reliever is successful in one of these, and leaves without losing the lead for his team, it's not a save, it's a hold - and doesn't count toward blown save percentage. So I'm not convinced save percentage measures the overall effectiveness of a bullpen either.

It's also significant that not all saves (situations or opportunities) are created equal. Most obviously, a one-run lead and a three-run lead in the ninth both get you a save, but blowing the latter is obviously indicative of a worse performance. And no team in the National League has been involved in more games decided by one run than the Diamondbacks' 27. It wouldn't seem a stretch to consider that this excess of close games would result in a greater number of "tough save" situations, with smaller leads than those enjoyed by most teams. Unfortunately, there are no stats that directly indicate the average margin of save situations.

However, there is something similar. Leverage Index (LI) was created by Tom Tango, to measure how important a particular situation is in a baseball game - it depends on the inning, score, outs, and number of players on base. The average situation is 1.00, with more important situations ranking higher. 10% of all real game situations have a LI greater than 2, while 60% have a LI less than 1: multiply the LI by 80, and you'll get a fan's average heart-rate in that situation. :) There are various numbers you can use to examine a player's work: LI when he enters, when he leaves, or the average value across all the time he was on the mound.

What I note is that, overall, the Diamondbacks' relievers have been working higher-leverage situations than anyone else in the league, and it's not even close. Their average LI is 1.33, with the second-place Reds all the way back on 1.19 - as some indicator, they're closer to the Rockies in 10th, than they are to us! Or, put another way, 35.5% of relief appearances were what Baseball-Reference.com calls "high leverage", with the first batter faced having an LI of 1.50 or higher. That was true for 39.8% of our appearances, third-highest in the league, and the raw number (88) is #1. Given this larger volume, a greater number of blown saves seems kinda reasonable.

Another way to look at relief pitcher performance is by the impact on Win Probability. Overall, Arizona is doing pretty well, with our bullpen having accumulated +309% over their appearances to date, third-best in the league. That's in stark contrast to the teams with similar numbers of blown saves to us, such as the Dodgers (-385%) and Cubs (-370%), who are last and second-last respectively. Again, blowing a one-run lead has a much smaller impact on W{ than doing so for a three-run lead, and again, this would suggest that the overall impact of the bullpen has been positive, more in line with their ERA than the blown saves.

Fangraphs has a pair of stats that come off this: Shutdowns and Meltdowns. You're all familiar with the graphs of Win Probability that we have in every recap? A relief pitcher is credited with a shutdown, if he improves his team's WP by 6% or more; conversely, a meltdown is when a reliever gets a WP of -6% or less. This helps work around some of the problems with the save statistic, such as the fact that it can only be earned at the very end of a game. Shutdowns and meltdowns take the situation into account, but can be awarded to any reliever. The following chart has the numbers by team, as well as the SD/MD percentage and some of the other numbers discussed above.

Again, this seems to show the D-backs bullpen in a decent light, with a ratio of shutdown to meltdowns that's in excess of 2.5. But what's also worth noticing is the individual performances by the team. Here are the same numbers, broken down by each relief pitcher who has seen significant use this season.

Ziegler has been our best reliever by WPA, courtesy of an excellent shutdown rate: of course, eliminating inherited runners with a double-play is certainly good for Win Probability! Bell and Hernandez have both seen the same degree of high-leverage situations and overall, their performance has been similar. Hernandez has a better shutdown percentage, but when he melts down, it's incandescent: he owns two of the three worst meltdowns by an AZ reliever this year Conversely, Collmenter's WPA numbers are heavily reliant, to the tune of +94%, on his two marathon relief outings, easily the shuttiest shutdowns of the season for us.

With Putz apparently due to come back shortly, it's perhaps worth noting that his struggles are perhaps a result of a very high LI, giving him virtually no margin for error. And when I say "high," I mean it's the highest LI of any reliever with 10+ IP in the majors this year. Only one of ten save opportunities into which he was inserted this season involved a three-run lead, It'd be nice if, when he comes back, the offense helped out and provided a few "easy saves" for his consumption!

[All stats exclude Wednesday's game]