Through the first 6 innings of last night’s game, it seemed like one of those nights where the Phillies just had no shot at winning. Cliff Lee was pitching well enough, but, as has been the story too often this season, the bats were silent. Following a John Buck single and Mike Stanton sacrifice fly in the bottom of the 6th, the Marlins led 3-0 and were sitting on a victory that was 91% assured, per win expectancy. Pack it in, savor the off-day, be thankful that the Phils didn’t get swept. In the very next inning, though, a Phillies rally was forged, as it seemingly so often is, by an improbable collection of hitters, and beginning with an opponent’s mistake. With Raul Ibanez on second, Gaby Sanchez was unable to handle a soft grounder from Dane Sardinha, putting men at the corners for pinch hitter Pete Orr. Orr fouled off four pitches and doubled on the sixth pitch of his at-bat, scoring Ibanez. Ross Gload, hitting for Lee, contributed an RBI ground out to bring the Phillies within one run. Victorino added a solo shot in the 8th to tie the game, and, all of the sudden, the game seemed very winnable. Then this happened:
If you were on Twitter, you probably saw the shockwave of disbelief that emanated forth. Likewise, I’m sure quite a few coffee tables in the Delaware Valley were pounded when the broadcast returned from commercial and Kendrick was perched on the mound. In the end, Kendrick got lucky (again); he allowed two baserunners but escaped the inning with a double play off the bat of John Buck. That does nothing to excuse the move, though, and it was far from an isolated event. Charlie Manuel has consistently used Kendrick in the most inexplicable of circumstances since the season began. Observe:
The above graph shows the leverage index for all of Ryan Madson and Kyle Kendrick’s batters faced in relief, sorted highest to lowest. As a brief refresher, or in case you are unfamiliar, leverage index is a sort of quantification of the importance of a given scenario, dependent upon the score, the inning, the base situation (runner on 1st, bases empty, bases loaded, etc.), and the number of outs. The higher the leverage index, the greater the change in win expectancy that can occur as a result of the plate appearance in question. The average is 1. By way of example, when John Buck batted for the home team in a tie game in the bottom of the eighth, with the go-ahead run on 2nd, the leverage index was 3.59 — quite high. With one swing of the bat, Buck could have radically altered the Marlins’ probability of winning the game. In the top of the 1st, when Placido Polanco batted with the bases empty and 2 outs, the leverage index was 0.40 — quite low. Since it was a scoreless tie and very early in the game, there is not much Polanco could have done to significantly change the probability that either team would win. You can “feel” leverage index as you watch the game. When you think to yourself, “this at bat is a huge moment,” the leverage index is likely very high. The number itself is just a way of objectively measuring it.
Anyway, what this graph shows is that, amazingly, the twenty-one most important scenarios in which Kyle Kendrick pitched in relief so far this season were all of a higher leverage than Ryan Madson’s top twenty-one. You’ve already seen this graph, but you don’t know it. Every time Kyle Kendrick has taken the mound, every time Charlie Manuel has produced some inscrutable quote about his bullpen strategy, every time Ryan Madson has been left sitting in the dugout, waiting through vital spots in the game for the arbitrary set of circumstances that constitute a save opportunity to occur, a piece of this graph spontaneously manifested on a higher plane of baseball existence and was burnt directly into your subconscious. Here it is now, in all of its terrible glory, for you to behold. Like any good blogger, I will say what need not be said: this graph should not be. If Charlie Manuel were properly leveraging his relievers, using the best ones for the most important points in the game, this graph would be flipped, and there would be a wider gulf between the lines. Instead, Manuel is electing to forgo his tire-chained Land Rover in favor of a 1979 Toyota Tercel on a blustery day in Lake County, Minnesota.
Don’t be fooled by Kendrick’s 1.42 ERA. We’ve seen enough warning track fly balls and first-pitch-swinging bailouts to know that he’s getting tremendously lucky. Kyle has a .213 BABIP in relief appearances this year. Kendrick’s SIERA is 6.56 — that is, given his current profile of walks, strikeouts, and batted ball types (flyballs, groundballs, popups) you would expect his ERA to be 6.56. SIERA is a better predictor of future ERA than ERA itself. The difference between Kyle’s SIERA and his ERA, Matt Swartz notes, “seems to be the biggest difference in calculable history of SIERA (2003-10) for anyone with 15 innings or more in a season.” The upshot is, Kyle Kendrick is going to come crashing down. Soon. To an extent not commonly seen. And when he does, the Phillies and their fans will pay dearly for Charlie Manuel’s decision to put him on the mound in some of the most crucial situations that the team faces. Add it to the list of factors working against the team sustaining its current pace.
While we’re looking at graphs that should not exist in their current form, how about this one:
J.C. Romero, re-signed on the cheap, ostensibly for his (overstated, in my opinion) effectiveness against left-handed batters, has faced 22 right-handed batters and 12 left-handed batters so far in 2011. In 2010 he faced 74 righties — 70 or so more than he should have. In fact, as David Hale tweeted, right-handers have accounted for more than half of Romero’s workload in each year since 2008, save 2010, where they still constituted 43% of his opposing batters. If Romero’s staggering futility against right-handed batters needs restating, allow me to quote myself:
For some reason that Charlie Manuel either hasn’t been asked about or hasn’t bothered to explain, an absurd 43% (74 out of 171) of the batters that J.C. faced in 2010 were right-handed — and Romero is distinctly, astoundingly awful against them. By win probability added, Romero cost the Phillies nearly two-thirds of a win in his efforts against right-handers, while he gained them almost half of a win against lefties. At least by that measure, allowing him to face so many right-handers entirely canceled out his success against lefties, and then some. Right-handed batters posted a .452 on-base percentage against Romero in 2010, despite only a .231 batting average, so you can guess how they were getting on-base; Romero walked an abysmal 11.85 righties per 9 innings, compared to a 5.27 K/9. Pitching against right-handed hitters, Romero’s usage of his fastball dropped about 28%, mostly in favor of his slider, which he struggles mightily to control. He threw 54 sliders to right-handed hitters in 2010, and just eight of them resulted in a called strike. A sizeable majority, 63%, were taken for balls. He induced only three swinging strikes and four fouls on the pitch.
For all that’s been said about this team — the below average offense, the easy April schedule, the injury issues — it is still one that is well-built to compete. But to overcome those sorts of obstacles, wins need to be squeezed from the margins of games. One way to do that is to use the bullpen as optimally as possible. The Phillies must match the pieces of their arsenal with the scenarios that are best fit to their talents and deficits. Charlie, old school baseball mind that he is, manages his bullpen almost always on the innings level, with roles defined in a way that has little to do with the hitter, the importance of a given situation, or the platoon matchup. His bullpen principles are simply not granular enough to be tailored most beneficially to the scenario at hand. Back before opening day, for Crashburn Alley’s Phillies blogger roundtable, I wrote this:
I actually find this pen to be acceptable, if it were used optimally.
This means that Romero can’t face 74 right-handed batters again, as he did in 2010. In fact, it would be preferable if he didn’t face any at all, although I suppose a few over the course of the season would be inevitable. It means that Baez should be banished to mop-up duty only, and Kendrick restricted to righties (who’ve hit just .258/.303/.397 against him in his career) in low to medium leverage situations. It means Contreras should log a lot of innings if he stays healthy, and preferably most of the high leverage situations that aren’t pitched by Ryan Madson.
I acknowledged later in my response that, knowing Manuel’s style, I had no expectation that things would happen that ideally. But I did not imagine that, one month and change into the season, the bullpen use would have unfolded so exactly at odds with what I was describing — almost perfectly bad. All things considered, the outcomes for the bullpen have not been that bad, but that’s not something that is sustainable. If these trends continue, the Phillies will be adding a self-created handicap to an already substantial list of challenges they face.