A Quick Defense of Matt Williams
On Saturday night, Matt Williams removed Jordan Zimmermann with one out to go, trying to preserve a 1-0 lead. Drew Storen entered immediately gave up a single to Buster Posey and a double Pablo Sandoval, tying the game at 1-1; the Giants would go on and win in 18. Williams was heavily criticized in the aftermath of the game for taking a pitcher out who was, at that, throwing a shutout, especially given that he had only thrown 100 pitches on the night.
But let’s just look at the numbers here for a second. We’ll look at career and 2014 batting against both pitchers.
| Career | BA | OBP | SLG |
|---|---|---|---|
| Zimmermann | 0.249 | 0.292 | 0.383 |
| Storen | 0.224 | 0.289 | 0.330 |
| 2014 | BA | OBP | SLG |
|---|---|---|---|
| Zimmermann | 0.244 | 0.277 | 0.354 |
| Storen | 0.210 | 0.262 | 0.278 |
On a per-batter faced, Drew Storen has been a more effective pitcher than Jordan Zimmermann. This shouldn’t be a big surprise, since good relievers are almost always the hardest guys in all of baseball to hit, and Storen is a very good reliever.
And as you’re probably sick of reading about by now, pitchers get worse the more often they face the same hitter within the same game. Here are Zimmermann’s career splits by times through the order:
| Split | PA | BA | OBP | SLG |
|---|---|---|---|---|
| 1st PA | 1,304 | 0.226 | 0.273 | 0.355 |
| 2nd PA | 1,291 | 0.256 | 0.293 | 0.389 |
| 3rd PA | 982 | 0.262 | 0.313 | 0.407 |
| 4th PA | 81 | 0.338 | 0.363 | 0.468 |
The first time Zimmermann faces a hitter in a game, he’s lights out. Second time, still pretty good. Third time, he’s roughly league average. The fourth match-up has been a disaster for him.
And yes, Zimmermann was throwing the ball well on Saturday night, but you only get to face a guy a fourth time through the order if you’re pitching well, so that entire data pool is essentially comprised of performances against Zimmermann late in games in which he had already performed at a high level. 81 plate appearances is of course a small sample, but both the league-wide and Zimmermann-specific trends are clear; his performance declines the longer he stays in, and by the time a hitter has already faced him three times, they hit him pretty well.
Even if we cherry pick the numbers, we can’t come up with a scenario where the expected line against Storen would be worse than against Zimmermann. Storen’s career .273 wOBA allowed is better than Zimmermann’s career-best .280 wOBA allowed this year, so even if we give Zimmermann credit for his 2014 performance and still hold Storen’s entire career line against him, Zimmermann still loses out. Add in any kind of times-through-the-order penalty and it ceases to even be remotely close.
Storen was the right call. It didn’t work, but that doesn’t make the move a mistake.
Dave is the Managing Editor of FanGraphs.
100% agree, and really dont understand the backlash
I do, and easily: there are a LOT more voices following baseball now in the playoffs than followed thru the regular season.
For example, the voices on Fox, who are, pick one:
A) negligently ignorant about what they’re describing (unforgivable, given all the available resources; and even more unforgivable, Ron D’uhling);
B) wilfully ignorant about what they’re describing, deliberately dumb it down for the presumed wider audience of presumed doofus ignoramuses begging for lazy narratives;
C) both.
I pick C.
i’m sure that move will somehow turn out bad for the nats!
Absolutely agree, but just a random comment/lament: it’s been a tough post-season for doing the sabermetrically right thing to do it seems!
That’s a fallacy. Just because the Royals are 3-0 does not mean that suddenly everyone should try to build a team like the Royals. Regression to the median is always expected.
Whathehell does the Royals being 3-0 have to do with regression to the median?
Mean. It’s ‘regression to the mean’ or ‘mien’. Median isn’t the same thing that all; it’s what you can run into if you’re not paying attention at the wheel.
Or, if you really want to sound like you know what you’re talking about, refer to the R-squared (correlation of multiple co-efficients) or r-squared (simple correlation of co-efficient).
Seems when managers have left their starting pitchers in longer than sabremetrics counsels, those starters have gotten rocked.
Of course, subsequently going to Thornton for Sandoval was also the right move.
I’m not sure any MLB manager makes it, though, having had his closer throw just one [good] pitch.
I don’t really understand the “multiple-time-through-the-order” thing. It seems so situation dependent. I don’t think you can say a logarithmic .100, .200, .300 can be always a benchmark rule. What is a pitcher is dominant but just slightly less dominant in the final few innings? Like a virtual .050, .060, .070?
The research done on the topic indicates there’s little reason to think a pitcher is “hot” during a game. That is, there’s no reason to think include how a pitcher has done so far in a game will affect his future performance. At least, it’s less indicative of future performance than a larger sample size.
It seems like the entire sample of Zimmerman’s is too small to predict much of anything. I get the TTO argument, and your main argument is probably correct, but it seems disingenuous to marshal a few thousand PA for an ad hoc argument rather than just pointing to a universal truth borne out by all starters’ stats.
It’s in part to head off the “but this is JORDAN ZIMMERMANN” response.
Cameron did both. He pointed out the universal truth (starters get worse each time through the order) and the stats for the individual pitcher. Not sure what else you’re looking for.
Anyone catch baseball pundit Mark Mulder and his fileting of Williams after the game? Mulder was eager to remind us all that he said at the time the move was made that it was a huge mistake, a/k/a “I told you so.” What a dick.
If Williams had chosen to stick with Zimmerman, and Posey promptly launched one that landed on the Capitol dome, the savages would be calling for Williams’s head too. There would have been the endless palavar of “Doesn’t he know how much Zimmerman struggles the fourth time through the order? Doesn’t he know Zimmerman has never been a strong finisher? Williams has three closers to choose from to get one lousy out, and instead he sticks with a tiring starter with a history of elbow problems?”
It’s easy to do well on a test if you know the answers ahead of time. Baseball doesn’t work that way.
+1. Although you forgot “didn’t he see how hard Posey has been hitting the ball of Zimmermann tonight?” and “didn’t he notice that a guy with 1 HR in 287 career trips to the plate came within less than an inch (on the bat) from hitting a second deck HR?”
Yeah, it’s pure carping and second guessing from the back seat managing crowd.
!
Mulder having been a starter, he’s in part saying “he should’ve left ME! in!!! ME!ME!ME!”
If Storen was the right move, why did Williams not put him in to START the 9th?
If it is the right call to put in a great reliever, why do so many managers wait til the starter gets in trouble? Uncle Buck seems to be the best at putting great relievers in the game at the right time, regardless of whether or not the starter is “pitching well”. Maybe Storen still blows the lead, but it is much easier to prevent a run when you start with the bases empty than it is with a runner on first.
Actually, I’d rather start with 2 outs and a man on first, but each to his own.
Very good point.
The difference is that was a good result… so result =/= process (apparently) doesn’t apply.
It only gets aired out when people don’t like the result.
Williams screwed up – he either should have never had Zimmerman pitching in the 9th (if you want to stick to the 4 times through the lineup reasoning) or had him finish it (still effective/pitching well)
It’s funny, because I’ve seen a LOT of bitching by commentators and fans that Williams managed by rote, on auto pilot. But if he’s on new school autopilot, he brings Storen in to start the 9th. And if he’s old school autopilot, he lets JZim try to get Posey. And it’s not like using his closer that way has been his practice this season.
Love it, lump it or hate it, the move wasn’t a reflex/push-button move.
Considering the usage of Storen, you are correct. If he wanted Drew, he should have used him to start the 9th inning. It has traditionally been Storen’s usage to have him come in at the beginning of an inning, while Tyler Clippard was always more likely to enter the game in the middle of an inning. I remember Davy Johnson remarking on how it took Storen longer to get ready than Clippard justifying some of his pitching usage.
That being said, regardless of who you use, Thornton needed to be up and ready for Panda.
Following this up with some data:
It seems my recollection as to who comes in with inherited runners recently is incorrect, and was based on their usage their first couple of years as Nats. However, Davy Johnson appears to have been right. SMALL SAMPLE ALERT:
Storen: 18/53 inherited runners scored
Clip: 4/24 inherited runners scored
That is 2012-2014, if you want to just focus on this year
Storen: 10/24
Clip: 0/6
I’m GUESSING that Williams was thinking, Ach, Buster was due against anyone. And maybe so; but Sandoval’s splits between batting lefty and righty scream for going to a left-hander there.
And don’t forget Bochy boned it too: Posey’s the slowest regular position player the Giants have, and they still had the fastest player on their roster, Gary Brown, a true burner, sitting on the bench, plus if Sandoval had only driven in one, they still had their very good rookie back-up catcher Susac on the bench to replace Posey.
In sum: regardless that Williams was correct in removing Zimmerman before Posey, Williams shouldna been in the position to be removed then in the first place, Williams blew it in not bringing in a lefty to force Sandoval to his far weaker right side, and Bochy blew the chance to score the go-ahead run from first. Not exactly a stellar sequence for professional in-game baseball management.
It seems to me to be at least as appropriate to harangue Williams for not using Storen to start the inning as it is for bringing him in with a runner on base.
None of the above takes into account that Storen’s last time pitching in the playoffs he had one of the biggest meltdowns in the history of playoff baseball.
A meltdown for which he received death threats and basically resulted in him losing his closer job.
Given that this was the case, it would have behooved Matt Williams to refrain from bringing him into a game with the Giants’ best hitter up and the tying run on base unless absolutely necessary.
Those that are criticizing the decision are taking into account that Storen was likely to be exceedingly nervous in his first foray back into the playoffs. Accordingly, it made sense to keep the guy in who had barely thrown 100 pitches and had retired 20 straight batters.
These players are humans, not automatons.
Storen made good pitches, which tends to debunk your “exceedingly nervous” theory. The Giants just hit them.
Posey was hitting .488 in the location of the pitch that he hit.
http://espn.go.com/mlb/player/hotzones/_/id/30112/buster-posey
The Sandoval pitch was in a location where he bats .389 or .290 vs. righties (depending how you look at it) Though to be fair Sandoval is tough almost everywhere vs. righties.
http://espn.go.com/mlb/player/hotzones/_/id/29212/pablo-sandoval
Can you imagine the postgame press conference if Storen (or Zimmerman if he’d have been left in) had chosen to pitch Posey or Sandoval up or in and they had knocked it out of the park?
I don’t care what any heat maps show, anybody pitching to those guys in the 9th inning when they represent the winning run are going to try to pitch them exactly as Storen did. Which could be a knock against conventional wisdom I suppose. But it shouldn’t really be against Storen.
Small sample size, as ElJimador points out below
Exactly. If he’d have come in and thrown meatballs in the middle of plate or pulled a Rick Ankiel then that might point to nerves. Instead he threw 3 pitches right on the edge of the strike zone and Posey and Sandoval just happened to hit 2 of them for base hits.
Sometimes a pitcher make good pitches and get beat, just like sometimes a manager makes the right move and it doesn’t pan out.
Sorry the downvote was an accident.
As for your contention, the pitch to Posey was in the worst place possible when pitching to him.
The Sandoval pitch was marginally better.
Location matters based on the hitter.
@CM Storen threw Posey a first pitch 95mph sinker down on the outside corner of the plate, that is a good pitch to any hitter. Second, there are very few 95mph nibblers. Storen has a low walk rate but its a little much to expect someone that throws that hard to have pinpoint accuracy.
The pitch to Sandoval was also a 94mph sinker at the bottom of the strike zone on the outer half of the plate. But Sandoval will swing at and can hit just about anything.
Sometimes hitters hit, even when the pitcher makes his pitches.
Sorry CM, I’m not going to buy that the outside edge of the plate or the base of the strike zone are the worst locations to pitch to any hitter with power when it’s the 9th inning and they represent the winning run. Those heat maps are fun to play with but they should not be taken as gospel.
Do you have a citation for the death threat thing?
http://www.huffingtonpost.com/2012/10/13/drew-storen-death-threats-twitter-nationals_n_1963243.html
This is silly. If Storen’s so fragile that he can’t be brought in during the 9th inning of a 1 run playoff game, he shouldn’t be on the roster.
Did you see him today? It was obvious that he was still fragile until he finally got a couple of outs. He is likely fine now, but it is silly to think that his nerves weren’t taking a toll. He gave up four straight hits over two games before getting an out.
This type of thing is the difference between a manager like Buck and a rookie like Williams. Buck would have done everything he could have to get Storen’s feet we in a better situation. I am sure that is part of the reason Williams put him out there with a 4 run lead today.
If anything I have more (not a lot) beef against Zimm. He is only Panik, dont try to nimble that much and challenge him with the 94 heat. Its harder to get a hit than to make an out.
Yes, that was really weird. Panik has an ISO of .063 this season, let him hit the freakin’ ball!
By that point in the game, very very few starters have 94 mph left to go to, ESPECIALLY those who’ve undergone Tommy John surgery. I think you’re maybe mistaking Zimmerman for someone like Kershaw or Bumgarner, who SOMETIMES are able to go into the 9th with that sort of speed still in store. Rather, I’d suggest that one big reason why Zimmerman’s numbers through the rotation get worse is as the game moves on, the batters are sitting on what’s left of his speedball and crushing it.
Taken with no context, Dave has a good point and the numbers back it up. However, it’s the playoffs when context matters more than at any other time. Zimmerman had given up just 3 hits over his past 17 2/3 innings when Matt Williams pulled him. He had just thrown pitch number 100 and, while his velocity had fallen off some later in the game, he did flash 94 during the 9th. He also had held Posey to a .200/.250/.200 in 16 plate appearances. On top of that, Panik walked on a borderline pitch that had been called a strike earlier in the game (see August’s post from today). Deciding in that situation that you’d rather have Storen, who famously blew the series in his last playoff appearance, pitching with the game on the line over Zimmerman is inexplicable to me.
Sorry, but 16 PAs don’t mean anything. If you read Fangraphs, you should know better. Look at Zimmermann’s numbers the 4th time through the lineup, these PA’s all came in games where he was pitching well, the’re not good.
Looking at Panik’s PA, Zimmerman was nibbling outside to a batter with a .063 ISO, that’s just not super intelligent when you’ve got Posey and Sandoval right behind him.
It’s obvious you take out Zimmermann after the walk to Panik, and there are good arguments to be made that Williams should have taken him out earlier.
What is the sample size for Zimmermann’s 4’th time through the lineup?
81 PAs, just like it says up above. And zillions and kadrillions as to the general awfulness of the 4th time through the order, again basically all of which apply to pitchers who were ‘on’ such that they got through the first 3 times in shape such as to take ’em on again a 4th time.
“However, it’s the playoffs when context matters more than any other time”.
From a narrative perspective, absolutely. From a perspective of players’ skills, there’s very close to nothing statistical to back that up.
From a statistical perspective we also know that statistics are not very effective at predicting what will happen in 1 at bat. In a single at bat, factors that cannot be so clearly quantified statistically (such as focus, adrenaline, nerves), play a much larger influence on the outcome than they do on the outcome of an event repeated 300 times.
Shauntell – 16 PA in many scenarios means nothing but in looking at one pitcher vs. one hitter it certainly starts to tell you something. There are only so many different looks and different pitches a pitcher can throw. After 16 PA a hitter has seen them all. I’m not saying that it’s the evidence that makes my case irrefutable, merely that it is evidence that is worthwhile to consider.
I believe in The Book, they show that past performance of a particular batter vs. a particular pitcher have little (if any) predictive value. And they were using matchups larger than 16 ABs.
I have more of a problem with Williams sending Zimmerman out to start the 9th in the first place. That was the highest leverage situation of the game Zimmerman had encountered to that point. As Dave notes, Zimmerman isn’t great a 4th time through the order (like every pitcher). To the extent that managers want to match up their best relievers with the highest leverage situations, Zimmerman was probably the worst available option to pitch into the 9th.
The only argument I might make is that Storen was perhaps was perhaps the wrong reliever. I was thinking Clippard would have been the right choice given the different pitching styles between he and Zimmermann. It crossed my mind that Storen has a somewhat similar delivery to that of Zimmermann. I’m obviously nitpicking and I don’t know the difference in matchup numbers between Clippard and Storen, I just like Clippards different arm angle and Funky delivery in that situation. I know it’s not a great argument, especially given how good Storen was down the stretch. Either way they had to get Zimmermann out after that walk; he was worked up by the calls by the home plate umpire.
Whether it’s Sabermetrics, or old school baseball philosophy the answer would come out the same. Its not unusual or controversial for a manager to let his starter go out in the ninth if his pitch count is low enough, and then if he lets a man get on then pull him.
If you want to criticize Williams then you could fault him for not bringing out Storen at the beginning of the ninth instead of waiting until Zimmerman put a man on.
I was second guessing Bochy for not putting in a pinch runner for Posey, but I guess with 2 outs that was the right call to leave Posey in the game, even though they might have won in the 9th if they had a faster runner at 1st.
see grady little/ pedro call.
There is whole element to this that nobody is talking about. I don’t think the major issue is whether or not Williams should have pulled Zimmerman for Storen when he did but rather why did he not bring in a lefty to turn sandoval around and make him bat righty. Sandoval throughout his career has been a significantly worse hitter as a righty than a lefty and it seems to be getting more pronounced in more recent years. Below are the splits of Sandoval as a lefty facing righties and as a righty facing lefties for this year, the last 2 years, the last 3 years and for a career. (there are a handful of at bats where due to injury he bated left or right against the same hand which aren’t represented)
Sandoval splits
2014 wOBA (OBA/SLG)
wOBA as Lefty .359 (.363/.461)
wOBA as Righty .247 (.244/.319)
2013-2014 wOBA (OBA/SLG)
wOBA as Lefty .350 (.352/.453)
wOBA as Righty .274 (.288/.330)
2012-2014 wOBA (OBA/SLG)
wOBA as Lefty .349 (.350/.456)
wOBA as Righty .287 (.301/.352)
Career wOBA (OBA/SLG)
wOBA as Lefty .363 (.357/.493)
wOBA as Righty .309 (.317/.391)
As you can see he is definitely a way worse hitter as a righty (as a Giants fan I was elated when they allowed him to bat left by facing a righty). I think the right move may have been to leave zimmerman in for Posey and then bring in a lefty (Blevins, maybe even Gio Gonzalez) to face Sandoval rather than bringing in Storen for Posey and burning him on just one batter when you pull him again sandoval. Either way I know one thing: I personally would have DEFINITELY brought in a lefty to turn Sandoval around.
Concur with the writer. Bochy, by these standards, did a lot worse. He sent in a rookie with all of 7IP career, who had been hit hard the previous night, to replace for THREE outs a pitcher who was pitching better than Zimmermann and was at 72 pitches. If Williams is a bad manager, what does that Bochy make?
It worked, so all is forgiven for Bochy. If Werth’s fly ball had gone out of the park, then we’d hear your message. 😉
But it was obvious he had a different game plan for the 18th inning. He threw more off speed pitches, not just fastballs. And now him being a rookie is relevant? I thought experience didn’t matter in saber land?
Rookie= insufficient info about his performance. Perhaps this was already included in “7IP”.
http://www.fangraphs.com/blogs/is-the-next-k-rod-lurking/
Sometimes the best option is a rookie.
Agree with the article, but one component of the 4th time through the batting order that I’ve never seen addressed is that each new time through the order he’s facing the top hitters first, and clearly, from those numbers, most pitchers don’t make it to the 6-9 hitters for a fourth time, so the stats should be compared relative to the performance by lineup position. The point will still stand, I imagine, but it’d probably look less drastic
If Zimmerman was having an outlier of a performance, which indeed he was, why are we trying to explain the decision not to stick with him with expected stats? The second and third time through the order for the Giants against Zimmerman, they were held hitless.
Every time he gets to the 4th time through the order he is having an outlier of a performance. Pitchers will get pulled before there 4th time through the order unless they are having an exceptionally good day.
Walking Joe Panik in the ninth inning of a one-run game with Buster Posey on deck is a bad way to persuade the manager that you should stay in the game.
The Nats lost not because Drew Storen didn’t pitch as he had been pitching this season; they lost because they had two hits in the last ten innings of the game and are down two games to none going into today because they haven’t had any offense to speak of in the short series so far, aside from Rendon. Is it too bad that Storen gave up the run that tied it? Yes. Will we ever know what Zimmerman might have done had he faced Posey? No, we never will. He was very wild with Joe Panik and given that it was not a bad decision to pull him there. . .and in his last ten appearances, covering 9 1/3 innings, Storen had allowed eight hits, no runs, while walking none and striking out eight. Did Williams’ decision pay off? No, it did not. . .but that does not make it a wrong decision based on the information he had. This criticism of him is absurd.
According to the Fangraphs link, I’m not surprised they took out Jordan Zimmerman. I mean, the guy hasn’t played since like 1999! And it was only one season then for the Mariners. I’m surprised they even gave him the ball, given how well Jordan Zimmermann was playing…
Drink!
Sorry grammar police.
” Drew Storen entered immediately gave up a single to Buster Posey and a double Pablo Sandoval, tying the game at 1-1;”
Need an “and” or comma after the world entered.