Matt Harvey, Hansel Robles, and Hindsight
Going into the season, we expected the NL East race to be fairly boring by the end of the year, and as expected, there’s not a lot of drama left about the likely outcome; of course, the fact that it’s the Mets and not the Nationals running away with the race is a pretty big surprise. Led by a quality rotation and a surging second-half offense, the Mets have put themselves in prime position to get to the postseason for the first time since 2006.
Of course, this being the Mets, there is still plenty of drama to go around, even with a big lead over the second-place Nationals and just two weeks left in the regular season. Lately, that drama has come from the team’s handling of Matt Harvey. It started off with a public disagreement between Scott Boras and the team about whether doctors recommended or required a 180 inning limit for Harvey this season, with Harvey initially appearing to side with his agent, but then coming around to the team’s side of things, stating that everyone is on the same page about his usage over the rest of the season. With Harvey quickly approaching that 180 inning threshold, but the Mets also wanting to retain the ability to use him in the postseason, the Mets are now shortening his remaining regular season innings in order to reduce the amount of stress his arm takes in his first year back from Tommy John surgery.
That limitation was on full display last night, when the Mets and Yankees met up on Sunday Night Baseball. With ESPN’s cameras showing the game across the country, Harvey dominated the Yankees, allowing just one hit and one walk in his five innings of work, striking out seven of the 18 batters he faced on the night. And with the Mets holding a tenuous 1-0 lead entering the sixth inning, Harvey was replaced by Hansel Robles, who proceeded to give up five runs in the sixth inning; his fellow relievers would give up six more, and the Yankees eventually won 11-2.
Predictably, the reaction to removing Harvey after just 77 dominating pitches is not a positive one this morning. Here’s Anthony Rieber from Newsday, for instance.
Harvey threw five innings. Gave up one infield hit. Walked one. Struck out seven. It was the Dark Knight at his best. His last pitch was 95 miles per hour. He made the Yankees look like minor-leaguers.
None of that mattered. What you saw with your own eyes on the baseball field didn’t matter. What the Mets needed as they try to nail down the National League East against the suddenly showing-late-signs-of-life Nationals didn’t matter.
All that mattered was a line on a chart somewhere that said Harvey could only go five innings — not six, good heavens not seven — because of some unproven benefits that might come his way in the postseason, or next year, or after he leaves the Mets as a free agent following the 2018 season.
Ridiculous. Arrogant. Unfortunate.
New York’s papers are filled with this same kind of criticism this morning; since Robles and the rest of the bullpen struggles, removing Harvey was clearly stupid, given how well he was pitching at the time. Except, it wasn’t, and more often than not, it’s probably going to work out just fine for the Mets.
First, let’s present a few pieces of actual facts. Here are the opposing batters lines against Matt Harvey this year, based on how many times they’ve faced him previously in that game.
| At-Bat | BA | OBP | SLG | OPS |
| 1 | 0.191 | 0.242 | 0.320 | 0.562 |
| 2 | 0.214 | 0.259 | 0.308 | 0.567 |
| 3 | 0.253 | 0.291 | 0.419 | 0.710 |
The first and second times through the order, hitters have done next to nothing against Harvey this year; he’s been nearly as dominant as he was pre-surgery. Beginning the third time through the order, however, Harvey has been decidedly mediocre, and this is true of most every starting pitcher in baseball. The times-through-the-order penalty is a well established effect that has been documented countless times, and applies to everyone, even pitchers who are throwing really well early on.
Despite the temptation to buy into the “he was throwing well, thus he would have continued to throw well” logic, the data simply refutes the idea that we could look at Harvey’s early-game dominance and presume that the Yankees would have continued to struggle against him simply because he pitched well the first five innings. Starting pitchers perform much better earlier on in the game than they do in the middle and later innings, and we simply can’t take their early-game performance and extrapolate it forward into the later innings.
Then, there’s the little matter of the fact that Hansel Robles has actually been quite good for the Mets this year. He’s struck out 28% of the batters he’s faced this year while allowing roughly average walk and home run totals, and while it’s probably not predictive of anything, his .240 BABIP suggests he’s at least not just throwing the ball down the middle to try and maximize his strikeout rate. Even after last night’s meltdown, opposing batters are hitting just .190/.267/.379 against Robles this year. To illustrate the point, here’s opposing batters lines against Robles this year, and against Harvey the third time through the order.
| Comparison | BA | OBP | SLG | OPS |
| Harvey, AB3 | 0.253 | 0.291 | 0.419 | 0.710 |
| Robles | 0.190 | 0.267 | 0.379 | 0.646 |
Yes, last night, Robles performed poorly, and it was juxtaposed against Harvey’s early-pull, making the narrative an easy one to sell. But thinking that taking Harvey out after facing 18 batters — meaning he would have begun the third-time-through-the-order with the next batter he faced — and replacing him with Robles made the Yankees more likely to put up a big rally is unsupported the evidence. A tiring Matt Harvey is just not much more (if any more) effective at getting outs than even just a reasonably decent middle reliever. And there’s plenty of reasons to think Robles is a reasonably decent middle reliever.
Last night’s results made for a very easy Monday morning story. For everyone who wants to rail against innings limits and modern pitcher usage, there is no better time to get on the soap box than when a starter is throwing well, is removed for an inferior-talented reliever, and then that reliever immediately gives up the lead. The problem is that the same soap box isn’t revisited when tiring starters are left in to face hitters a third or fourth time within the same game, and that lead disappears before the bullpen is ever called upon; this happens all the time, but the story told then is that the pitcher simply failed to do his job, rather than that the manager failed to recognize that going to to a reliever would have provided a better opportunity for a good outcome.
It might be frustrating to watch, but the reality is that if you’re going to limit a pitcher’s workload, the most rational way to do it is to shift as many innings as he can throw towards the beginning of games, when he’s facing hitters only two times each, rather than skipping starts; these shortened starts allow Harvey to still pitch when he’s most effective and rest when he’s likely to be least effective. If you’re going to enforce an innings limit, this is the best way to do it, even if you’re going to take heat for it on the days when the bullpen doesn’t live up to Harvey’s early-inning standards.
The key is to remember that Harvey probably wouldn’t have lived up to those standards either. The choice the Mets made wasn’t to take out a guy who going to continue to be unhittable and replace him with a batting practice machine. In reality, the Mets took out a starter who has been roughly an average pitcher the third time through the order this year, and replaced him with a perfectly solid reliever. It didn’t work this time, and it happened on a national stage, but it’s certainly not ridiculous to realize that letting Harvey pitch deep into games has only a marginal benefit versus handing those middle innings to relievers who are likely to perform at a similar (or better) level.
If you want to debate the merit of managing workloads in general, that’s another story, but given that the Mets are going to manage Harvey’s workload — and with the fact that he wants his workload managed, they don’t really have a choice — giving him shortened outings is probably the best way to go about this. Despite a highly-publicized failure last night, concluding that the Mets plan is “arrogant” or “unfortunate” is simply ignoring the pertinent facts.
Dave is the Managing Editor of FanGraphs.
As a mets fan that thinks Harvey’s sideshow has detracted from our quest for the playoffs, this stinks in so many ways. However, we’re also forgetting the fact that we committed two awful errors in that inning to open the floodgates.
I think people are also forgetting that Harvey may be having the best comeback season ever from TJ. With all the TJs that team has had lately, you’d think people would be more understanding. Abusing Santana to get that no hitter basically ended his career. Harvey has already missed a season. Wheeler as well. The Mets are basically in the playoffs, why ride these guys hard for a meaningless game? As a Cubs fan, I’ll take him if you guys are tired of the sideshow.
big difference between pitching 77 pitches vs 134 pitches. also a difference between a shoulder capsule surgery vs a tommy john surgery….
serious apples vs oranges.
First of all, the game is not meaningless. The Mets have a while to go before clinching. Second, the Mets have not ridden Harvey hard this year. They have kept his pitch count low and tried to limit his starts. They likewise did not let him pitch last year (supposedly against his wishes), allowing him 17 months to recover from TJ surgery. Third, the lack of “understanding” comes from the fact that, until a few weeks ago, there was never a discussion of a hard innings limit and Harvey’s actions (complaining about not starting the first 6 games of the season, complaining about a six man rotation, complaining about being pulled when he was pitching well) were completely inconsistent with someone who believed he had a hard limit. Thus, it appears that Harvey is full of shit and that he only cares about making money and not helping his team. (He has never complained that he feels fatigued, so it is not a matter of him being forced to pitch against his wishes) Having Scott Boras speak for him hardly dispelled the notion that Harvey was a me-first guy intent on benefitting himself to the detriment of his teammates, who were fighting for a playoff spot (and home field for the first series). As a Mets fan, if you are offering Bryant, you can have him.
yeah, I don’t want him that badly…
I may have not written that clearly. I wasn’t saying that they have been riding him hard, just that they’re up by 6 games with two weeks to go. Save it for the playoffs. I’m applauding the team for not riding him hard. They’ve done a great job managing his usage which will allow the team to use him in the playoffs. As a baseball fan, I hope to keep watching him in the future.
Wait, db, you’re saying you think the Cubs would even think about giving up Bryant for Harvey? And you’re also implying that you previously wouldn’t have made that trade? HAHAHAHAHA
Bryant is better, younger, under control for longer, and less of a health risk
He’s more valuable and it’s not even close
Kevin. That depends on who you thing Bryant is. Yes he is young and might be the next trout. But there are lots of red flags. For example the high k rate that got higher in the 2nd half, the high Babip with is supporting his .260 average and his fielding at 3rd which most scouts believes he will move off. If Bryant end up a .240 hitting rf. he is really Adam Dunn. That is not even remotely close to a top 5 sp
Kevin
I’m not saying what the Cubs think, and objectively Bryant is probably more valuable. Point is, I would trade Harvey for good value. I don’t think mets fans are. Early so willi g to trade deGrom, who does not seem to be a phony.
Seriously though, I hope you guys are enjoying his talent. Forget the sideshow… it’s been fun to watch him from Chicago. I’d love to have him here. You guys have had a great season and will be in the playoffs as well. I know the situation is frustrating, but I hope you guys are enjoying the season.
Dave, you write the following as the last sentence of your fine article:
Despite a highly-publicized failure last night, concluding that the Mets plan is “arrogant” or “unfortunate” is simply ignoring the pertinent facts.
But I don’t think Anthony Rieber of Newsday was referring to “the Mets plan” as arrogant or unfortunate but rather to Harvey, Boras, and the Harvey/Boras plan of limiting his innings as arrogant and unfortunate.
This article by Jayson Stark is really terrific as it points out that a better indicator of Harvey’s workload is the number of pitches he has thrown this year as the average per inning is quite low:
http://espn.go.com/mlb/story/_/id/13675877/are-new-york-mets-mismanaging-matt-harvey
I’m not generally a fan of ESPN articles but I thought this particular article was terrific.
I don’t have a comment, other than to say I’m just here to read all the petty criticisms of Dave. I’m sure you guys won’t disappoint.
Would it be better to have only praise for everything written here? Wow, I don’t get this comment at all.
Go back into prior articles and chats. Criticism is one thing, a couple of people have really gone over the top. It feels like some readers need Dave’s validation or they take it very personally.
I think skipped over or don’t understand what the word ‘petty’ means.
Petty according to you?
After all the negative press this story has been getting, it’s good to see some positivity. The Mets have actually handled this pretty well–especially considering the Mets tried to manage Harvey’s workload several times earlier this season and he publicly criticized the idea. The shortened regular starts is a perfect solution to a difficult problem, and the story shouldn’t be about anything other than a bullpen that this time wasn’t the best even though it’s usually pretty good.
First off, I think that even had Harvey stayed in the game, he would have been lifted before or during the 7th, and the Mets still would have lost. The Yankees are a better team and showed it during their head to head games this year.
Second. I am really starting to believe that Boras created this Harvey controversy in part as a favor to his Nationals clients, 6 of whom are big-name players.
Lastly, the articles on this site burying the Nationals and even floating a Cespedes MVP candidacy were very premature.
With 3 games head to head to finish the season, this race is far from over, and Harper is clearly the more valuable player.
IIRC, the articles on this site did not advocate a Cespedes MVP but instead reinforced the idea that Harper and the player who spent the whole year in the league deserved the prize.
I do not think had Harvey stayed in for the 6th and possibly the 7th the Yankees would have won. With Reed, Clippard and Familia, I think they could easily have held them off. I also think that without the handcuffs of artificial innings limits on Harvey, and with their starters including Matz optimally lined up, the Mets could easily beat the Yankees in a 5 or 7-game series.
“The Yankees are a better team and showed it during their head to head games this year.”
The Yankees proved they are better based on six head to head games? For a guy who comments on this site a lot, your ignorance towards basic concepts like small sample sizes and randomness is pretty stunning. The Phillies won five of seven against the Cubs this year. By your logic, the Phillies proved that they are the better team.
“Lastly, the articles on this site burying the Nationals and even floating a Cespedes MVP candidacy were very premature.
With 3 games head to head to finish the season, this race is far from over, and Harper is clearly the more valuable player.”
No shit, Harper is the most valuable player. There hasn’t been a single article on this site that has disputed that. The articles that centered around Cespedes and his MVP credentials were more or less driven by the fact that mainstream writers like Rosenthal, Morosi and Posnanski wrote pieces discussing it.
I’m not even going to even bother addressing the ridiculousness of your second paragraph.
Yeah, we know that the Yanks are better because they have a better run differential despite a tougher schedule!
Head to head doesn’t matter that much.
U mad bro ?
Chill out. Every, and I mean every, writer / commenter on this site falls back on small sample sizes to make their points when they need to. Every one of them.
The point I’ve been making from mid-April about the Mets is the same one I’ll keep making: the bullpen ex-Familia is not good enough to win a title.
The Clippard trade was great, but it’s not enough. The Mets have a problem in innings 6 and 7, and didn’t address it. Other playoff teams are in the same boat, but most of them have more potent offenses.
The Nats are such a team, and this race will be a lot closer than the 1.5% probability given to them to win the division, despite the plethora of down votes my comments get.
I’m not mad at all, I just call it like I see it. What I saw, along with 12 others and counting, was a whole lot of stupid in the comment you posted. And the division race is over. Even the Mets can’t blow this. Their next ten games come against the Braves (60-90), Phillies (56-94) and Reds (63-85). Even if they only manage to go 5-5 in those ten games, the Nats would have to go 8-2 just to be three down with three to play. That’s why the Nats division odds are at 1.5%.
“Even the Mets can’t blow this”
Funnily enough, I recall hearing something similar around this time 8 years ago…
I’ll admit to a little bit of stupid in my post, but I’d stand behind most of my comments.
Short series or not, the Yankees and Mets have played an entire season to approximately the same record. So saying this is like a 60-90 team sweeping a 90-60 team and proclaiming the 60 team “better” is BS. It’s effectively a playoff level series between 2 teams somewhere around 8th best in the league by record.
I think that they still would have lost if Harvey pitched the 6th inning.
The Boras thing is a bit conspiratorial, but why wouldn’t he ?
The Nats are not done, down votes don’t change that.
Let’s see what it looks like on Friday …
And considering that “he’s so hot right now” got 9 upvotes, maybe getting 12 and counting downvotes is not a bad thing.
I don’t see it. The Nats are currently in the middle of an eight out of nine run. They’ll probably need to win eight or nine of the next ten just to make the final three game series against the Mets to mean something. Then they’ll likely need to sweep the Mets. Again there’s a reason the Mets odds currently stand at 98.5% odds.
I agree that it’s unlikely, but I’d say the Nats could roll off 8 of 10 while the Mets go 5-5 with a 25% probability, and sweep them with another 25%. That’s about 6%: still unlikely, but 4x more likely than the straight up analysis would dictate.
Obviously I’m not assigning 50/50 probability to wins like a rational person, but I think momentum is an actual short-term thing in baseball, and I think the Nats have it right now, and the Mets do not ( and they’re also playing bad teams ).
It’s that damn Hansel!
He’s so hot right now.
I’m just glad this wasn’t another eugoogily for the Mets.
Hansel and Goeddel. A fairy-tale bullpen. They should trade for Grimm.
The analysis in this piece is what I hate about the Sabre take on everything. It dismisses the specific case in favor of the general trend. Even though SPs tend to be more dominant in the early innings, they often dominate an entire game (or close to it). Sure, just because Harvey was dominant thru 5 yesterday, doesn’t mean he’d have continued that way but without the contrived IP limitations he’d have gotten the chance to shut the Yankees down and given the pen a much needed break for a day.
Not having a shut-down bullpen, the Mets cannot ever go on cruise control handing the ball over to the pen in a tight game. If they had the Yankees or Royals or O’s pen, it’d be a little easier too sell.
Exactly, baseball is as much a mental game as it is a game of numbers. I love Sabermetrics and the advanced analysis that it allows of players, but any person who says the Mets were better off pitching Robles over Harvey has never played a game of baseball in their life.
Sure, the third time in the order pitchers get worse. But given that, wouldn’t teams yank their starter roughly in the 6th inning regardless of performance. Every hitter on the Yankees got a mental boost from seeing a guy who made them “look like minor-leaguers” sit down 2 innings earlier than he should’ve. That matters more than the numbers.
The Rays have relived starters after the 6th inning to great effect
“It dismisses the specific case in favor of the general trend.”
Is there any evidence for this specific case that the result should be expected to differ?
“Every hitter on the Yankees got a mental boost from seeing a guy who made them “look like minor-leaguers” sit down 2 innings earlier than he should’ve. That matters more than the numbers.”
So it can’t be quantified, but we’re certain it’s a larger effect than what we can measure?
We’re not “certain” of anything here. We’re using statistics to get an educated guess of what we can expect from any given circumstance.
I’m sure Harvey has had games where his stuff hasn’t been all there, and in those situations you might be expected to see the expected uptick in batting average, slugging, etc. every time through the order.
In this circumstance, Yankees hitters were batting 1-16(.065) against Harvey with seven strikeouts. His stuff was all there, he had something the Yankees could not figure out. I’m all for the stats but the notion that Robles was a better option than Harvey is absolute nonsense.
“We’re using statistics to get an educated guess of what we can expect from any given circumstance.”
Yep, except Dave is using data from thousands of PA’s going back years, including this year, and you’re using 16 PA’s to bolster your case. It’s true that Harvey might have pitched well for the next couple of innings but, as the statistics point out, the data is strongly in Robles’s favor. You’re simply electing to ignore that data because it doesn’t fit your preconceived notions. The educated guess, therefore, favors Robles.
Education on this subject would indicate that 16 PA’s is far too small a sample to provide any predictive power and any guess that leads one to believe that Harvey was the better option isn’t supported by the data and, therefore, isn’t an educated guess.
I’m not using the 16 at-bats as proof of anything, all I’m suggesting is that the data needs to be examined, how often does a pitcher who is this much on a roll fall apart the third time in the order, relative to a regular start?
Giving up 1 hit over 6 frames with 6 strikeouts means the pitcher likely has sharp command of their pitches, and pitchers in that situation are different than pitchers who have given up 5 hits over 4 innings or something to that degree.
I obviously can’t pin this down with a study but claiming that every single pitcher going through the order 3rd time adheres to the same expectancy is limiting what the statistics can tell you, and I think it should be looked at. Remember that a few years ago saber-metrics thought base stealing was 100% harmful statistically?
I am waiting for the first team to try this out as a general rule, but even more exaggerated than this. I’m thinking the game is headed where a starting pitcher will go no more than 4 innings, get a couple days off, then come back to pitch another 4 innings, etc. We will have a 6-man rotation where picher 1 pitches the first 3-4 innings of game 1, pitcher 2 pitches the next 3-4 innings (the rest to be covered by the bullpen), pitcher 3 pitches the first 3-4 innings of the next game, etc. This will reduce stress on pitchers’ arms by not having them throw a lot of pitches on a given day, and not run into the scenario where pitchers are going thru the order 3-4 times, when the stats show they are less effective (whether they are tired, or the hitters have their timing down better, or both). The only problem is teams would have to come up with 6 core pitchers to do this with, where teams are struggling to come up with 4th and 5th starts now. Also, do you put your #1 and #2 guys in the same game, or do you put your #1 and your #3 in the same game, and your #2 and #4 in the next game, etc.?
The “Sabre Take on Everything” is merely trying to establish the best baseline course of action. Of course, every situation is different, and the “Sabre Take” isn’t discounting that. Analyzing an individual anecdotal event without looking at the overall situation and numbers and making grand proclamations about what should and shouldn’t have happened doesn’t serve anyone, which is what Dave was saying above.
Take a six sided die. You can place a bet that will pay out on a 1 2 3 or 4. You place the bet and the number comes up a 6. That doesn’t make the bet a bad bet. That’s what these kneejerk narrative-driven appeal to emotion pieces are.
“Even though SPs tend to be more dominant in the early innings, they often dominate an entire game (or close to it). ”
Except more often, they blow it in the 6th or 7th. That’s the point. The bullpen is better.
“given the pen a much needed break for a day.”
It’s September. The Mets could stock 20 bullpen arms if they wanted to. There is no need to manufacture rest for the bullpen by leaning on the starters.
I addressed this. If this is the “progressive” view, then it could criticized as a poor decision every time a SP is left in past 5 innings (or the 2nd time around the order).
The defense for following the generalization is there if you want it. Even on this level of analysis, no two situations are alike and that’s before we get to the human element. As pointed out by others, Harvey was on extra rest, had thrown only 77 pitches, was having his way with the Yankee lineup, and had some added motivation of getting the team closer to clinching. I don’t think he’d been in that same spot before, so the general history doesn’t really mean much.
Yeah in a spot where your SPer is laboring thru 5 frames, 100 pitches and multiple base runners, then it’s easy to justify switching to a fresh arm. Last night it was very debatable however.
Grrroowl!!! Aaahh, meow….too sell or not too sell a bullpen. Me dunno. Me don’t have too worry about this in hockey.
I might agree with the article if not for the fact that Harvey was pitching on 11 days rest and had only thrown 74 pitches in his previous start. So he was especially well-rested, and any fatigue which might contribute to a falloff the 3rd or 4th time a batter faces a starter likely would not have been in play last night.
Further, Robles is a rookie pitching for the first time in a fully packed house on national TV.
Given all these circumstances, I choose Harvey over Robles for the 6th and 7th innings 10 out of 10 times. Even if Harvey had been on just normal 4-days rest, I’d choose him over Robles, assuming he had coasted for most of the game. Only if he had struggled and had some difficult innings would I have opted for Robles.
Regardless, Harvey has lost a lot of credibility with his teammates, fans, and the media — not solely because of the hard innings limits which are being imposed, but also because of the poor way he and Boras have handled the whole thing.
This is why people suck at probability… MotherMary, and Metsfan… you deride the general over the specific… because THIS time, Harvey would have shut him down. The general, as in, everything that has ever happened, shows he is far more likely to not keep pitching that well. But since we can’t PROVE it would have happened this one time, you can cling to whatever you like.
I am not saying Harvey would have shut them down had he continued. I am saying under the specific circumstances, I would take my chances with Harvey every time.
The statistics of Harvey facing batters the 3rd and 4th time through the order were not compiled when Harvey was on 11 days rest. And a rookie Mets reliever facing the Marlins in a half empty stadium is far different than facing the Yankees in a packed house on national TV. The pressure in the 2 situations is not the same.
If he’s feeling pressure because it’s his first time on national TV, doesn’t it make sense to get it out of the way now rather than the playoffs? Yesterday’s game was fairly meaningless – it helped the Yankees more than it hurt the Mets – so the last thing you want is for Robles to pitch a meaningful inning in October and show these jitters.
viva — This series had long been viewed as an audition and rehearsal for the postseason. It was talked about in the media. And to that extent, it helped the Mets narrow down choices.
But last night that was simply a silver lining to the cloud. The main goal was to win the game — especially since the division has not been wrapped up yet. Then once you clinch, you can rest people and experiment all you want.
Unfortunately for Robles, he just about flunked his audition.
I’m not proclaiming that Harvey would have done anything in particular. I’m saying that performance varies each time and circumstances are never the same, and so we shouldn’t be so rigid in making decisions according to what the general trends suggest.
If we take this to its natural extension, Harvey (and I guess most SPs) would never pitch past the 5th inning. But sometimes SPs do have special days and actually shut teams down for 8 or 9 innings.
I just posted a reply above where I make a point regarding your “natural extension” of most SPs not being allowed to go past the 5th inning. I think that day is coming.
And where will that end? Then the analytics will say that results 1st time around the order are better than the 2nd, etc….
FIP — You want the “general” as in “everything that has ever happened” before?
Here’s everything in Harvey’s career for how he’s pitched with various days rest:
Rest – BA/OBP/SLG/OPS
4 days – .225/.273/.368/.642
5 days – .226/.263/.303/.566
6+ days – .159/.257/.230/.467
He clearly has much better numbers when pitching on extended rest, which mitigates the stats on number of times through the order.
Again, combined with the fact Robles was a rookie in the most pressure-packed situation he’d faced to date, I would pick Harvey for the 6th every time.
You need to break that down by times through the order. Also how many starts does he have on 6+ days rest? 5?
It was a perfectly fine decision to go to Robles. Saying you’d go to Harvey every time just makes you look stubborn.
Adam — I don’t have the data to break it down by order. But the same argument could be made of the data that Dave presented — it needs to be broken down by number of days rest.
Suffice it to say there is clear evidence that Harvey pitches better on more rest, which would change the equation used in the original article. And again — couple that with the high-pressure environment Robles was faced with, and it practically begs to leave Harvey in. This is not being stubborn, it’s using common sense and wider set of information than Dave was considering.
I’ve read no article that stipulates that he has lost credibility of his teammates. The media has created the narrative, yes,perpetuated by Harvey and his agent (whom will remain nameless as to avoid his existence in this comment). Sadly, this knife has cut both ways. Met fans (and I’m one of them) don’t like dropping a game off their lead-we’ve been conditioned that this is not a playoff team until mathematics say so. With that being said, they did the right thing last night in the guise of “lose the battle-win the war”.
His teammates are not going to come right out and bash him. But read between the lines:
http://www.newsday.com/sports/columnists/anthony-rieber/innings-limit-for-mets-matt-harvey-turns-into-perfect-storm-of-failure-1.10869268
10 out of 10 times? So if Harvey gives up 5 runs in the first couple of innings you would keep him in no matter what?
I said “given these circumstances.” A 5-run lead changes the circumstances.
I’m skeptical of Dave’s general argument (for example, on the wild card) that any old fresh reliever is preferable to an ace going through the order a third time — this may be what the numbers show now, when aces regularly go 6-9 innings and relievers are rarely used for more than one inning, but if it became the norm there wouldn’t be enough quality relief innings to go around.
But Dave is right in this case: pulling a pitcher early certainly makes sense when there is an independent health reason to limit innings.
The problem I have is that the “independent health reason” in this case is not some concrete definable factor — such as pain, soreness, or injury. Harvey is almost 2 years removed from the actual surgery, and the “reason” in this case is simply precaution based on theory and speculation.
It would be interesting to see a team experiment a bit more with ‘sharing’ rotations spots, particularly with young pitchers or if they are coming off a major injury and need to have innings capped. It would require having six pitchers technically in the rotation, but four of them would essentially be pitching on normal rest, while the other two would be doing shared starts and shorter relief stints in between starts (you could theoretically do this with 2-3 of your rotation spots).
This analysis assumes Harvey was performing at a season average performance level during the game. Since he gave up 1 single in 16 AB and 2 base runners in 17 PA, then his OPS against was 0.180. I don’t think his skill during the game was close to season average unless you really want to attribute all of the improved results to luck.
With a 3rd time through the lineup penalty, his OPS against might have doubled to 0.360 or even tripled to 0.540, and still would have been better than Robles.
The third time through the lineup penalty really isn’t regression to the season average performance. If Harvey had given up 5 home runs and 10 hits in the first two times through the lineup, would you really expect his performance level to improve to 0.710 OPS against the third time through?
That’s not really the issue. The data shows that Harvey, like every other pitcher, gets measurably worse the third time through the order. In fact, your point about the 5 home runs only furthers this. Had that happened, Harvey would not have been allowed to continue pitching. If he makes it through the order 3 times, it’s more likely that he was pitching well. And yet he still performs worse in those situations.
I get it, Mets fans are paranoid about a collapse. I don’t blame them. But objectively, this was a regular season game with not a lot to play for. The Mets played bad defense and couldn’t put together enough offense against Sabathia, and the Yankees scored 11 runs. Not really worth it to send Harvey out for any longer than he was.
“unless you really want to attribute all of the improved results to luck”
I would. 17 PA tells us next to nothing. The chapter in The Book that deals with this shows that even pitchers who are pitching amazingly are prone to giving up a lot third (and fourth) time through the order.
Is there somewhere I can go to pull data on times through the order for individual pitchers? I’m genuinely curious, but I didn’t see it as an available “split” on the player pages here.
Try this:
http://www.baseball-reference.com/players/split.cgi?id=harvema01&year=Career&t=p#times
Harvey’s career splits:
1st PA: .191/.236/.303
2nd PA: .193/.252/.283
3rd PA: .241/.287/.361
Baseball Reference has them, but as I said above, I’m not too sure how relevant they are in this situation since those stats for Harvey were compiled for the most part under different circumstances than existed last night.
Those are convenient goalposts to set up, Metsfan.
Except that Baseball Reference recognizes those goalposts too.
Harvey on various days rest —
Rest – BA/OBP/SLG/OPS
4 days – .225/.273/.368/.642
5 days – .226/.263/.303/.566
6+ days – .159/.257/.230/.467
I’m (sincerely) not sure how much stock we can put into 88.2 career innings. Seems there would potentially be a lot of noise.
The numbers I gave are “everything that has gone before” and the specific 6+ days number is consistent with the others in showing a clear trend – the more rest Harvey gets, the better he pitches.
Moreover, only 406 plate appearances are included for the 3rd time around the order stat. That’s not much more than the 343 included in the 6+ days stat. Tell me, why would 343 PA include so much more noise than 406 PA — to the point that one stat is invalid and the other isn’t?
Weird that people aren’t making the opposite argument. That Harvey was going no more than 5 was pretty well known going into the game. Thus the issue actually came in bottom 4:
Mets get two men on with two outs up only 1 run. Harvey is at 60 something pitches, near where reports had the Mets wanted him to be. He is at most getting one more inning, and 7-8-9 was due up for NYY.
The Mets had Harvey hit, and he made the last out. A strong case could’ve been made for pulling Harvey THEN and trying to score additional runs. At worst, you save an inning for Harvey later, if you are in fact counting innings.
That’s a good point. And more in line with the “That Harvey was going no more than 5 was pretty well known going into the game”, is it possible that he pitched better in those 5 innings because he knew that he wasn’t going 100 pitches? It happens all the time with relievers, where they’re able to give max effort in short spurts. Maybe a similar thing happened with Harvey.
I did read earlier in the year that he was making an effort to pitch to contact more, and save his top fastball for when he needed it. His strategy this year has been about being more efficient. Knowing he wasn’t going past the 5th inning may have allowed him to throw that out the window, and worry about performance over efficiency.
In fact, MGL made this exact point on twitter. Well done.
These analyses of trying to apply information gleaned from trends averaged over a multitude of game situations to one specific situation are really quite misleading. There are many ways to get to an average, and every third trip through the order, or relief appearance is not the same. Harvey on the third time through the order may be no better than Robles on average, but that doesn’t mean we shouldn’t have expected him to be better on this specific occasion. To improve this analysis you need to consider the shapes of the distributions of expected outcomes, and not just the means.
I think – against the thrust of the article – that pulling Harvey probably reduced the Mets’ chances of winning the game. I think that a normal course of action would be to look for signs that Harvey was tiring (pitches up, missing his spots, etc.) and then react, rather than to pull him to avoid the 3rd-time-through-the-order effect.
But I think that the importance of this game, relative to future games that Harvey might pitch, is the important question. The Mets may have reduced their chances for winning this game, but they increased their chances for winning future, more important games, by pulling Harvey after 5.
Does the “times through the order penalty” have a split of its own? i.e. On days like last night when Harvey is dealing what are his 3rd time through the order numbers compared to days when he’s simply pedestrian?
You’d obviously have to develop a metric upon which to base it i.e. swinging strike rate? hard contact rate? k-rate? some combo? in order to classify which outings are dominant vs. pedestrian. But wouldn’t it be valuable if there were a predictive tool for understanding when the third time through the order penalty was likely to be more/less severe? Seems like it would have huge implications for bullpen management.
Obviously under ordinary circumstances Harvey pitches the sixth. Maybe even under some extraordinary circumstances like the playoffs they let him pitch. But the situation is one where they don’t want to push Harvey too hard in games that realistically don’t have that much impact. So they brought in Robles. No one is saying Robles is better than Harvey. The numbers suggest however that Robles could reasonably be expected to pitch about as well as a third-time-through Harvey over the course of one inning. We are left to speculate on how Harvey would have done, but it’s important to note that 1. Robles is not a bad reliever and 2. Robles was not helped out by his defense that inning.
In the back of every Mets fan’s mind is the voice of a Yankee fan who’s been giving him agita for many years. That Yankee fan’s voice spoke up loud and clear mocking the Harvey innings limit situation as it unfolded in a typically Mets-ish manner in the press. That voice cackled with anticipatory schadenfreude as Collins announced that the fifth was to be Harvey’s last inning. Then that cackling turned into full blown mockery and I Told You So’s nonstop for the rest of the game.
I’m a Mets fan. I know.
So when our sportswriters write these articles, they know who they’re targeting. It’s me, the Mets fan with that gd Yankee voice in my head.
Honestly, most Yankees fans don’t really care about the Mets one way or another. This is a rivalry that exists primarily in one direction. Personally, I root for the Mets in all games that they are not playing the Yankees. I want them to get to the WS this year and win it, so long as they aren’t playing the Yankees.
Just wish them luck on Eovaldi’s (and probably Tanaka’s) surgery, when they go after you for being too careful with your pitchers.
“I know that rushing Eovaldi back so you can brutally overuse him in the playoffs will definitely result in a World Series (in year where the Yanks aren’t even the best team in their own division) so it’ll be fine when he’s never the same pitcher again!”
I actually root for the Yankees in most circumstances, too, but some fair-weather Yankees fans in NYC are even more obnoxious than the bandwagoniest Bostonians.
Great article Dave and presented very well.
It would be interesting to see an analysis of the effectiveness of pitchers throwing 92-93 replacing starters, of the same side, throwing 95. As a hitter, it would be the same as a starter losing 2 mph off the fastball. While it may have been a good idea to replace Harvey, what other (lefties, harder throwers, etc?) would have been better to replace Harvey with?
But Robles throws over 95 mph with his 4-seamer, so there shouldn’t have been much of a drop-off (if any) from Harvey.
The Mets are in a difficult position with regard to Harvey. There are no clinical studies that have identified precisely what the risk of re-injury is based on number of post-Tommy John innings pitched, number of pitches thrown, etc.
Until there is, arbitrary innings pitched limits will be hotly debated for years by agents, players, doctors, and ballclubs.
All I know is the Mets are a better team when Harvey pitches. First and foremost, regardless of what may seem to point to the contrary, the Mets are intending to win each game to the best of their ability. Last night was no different except that they made a move to simultaneously temper Harvey’s usage while still maintaining their best opportunity to win the game sans Harvey (i.e., taking him out just before 3rd time through lineup). This article demonstrates that what happened yesterday was a good, well-informed decision by the Mets that just happened to result an undesirable (and I dare say unlikely) outcome.
I understand what the Mets did, but the fan in me wants Harvey fully unleashed. I’ve seen him pitch a few times, and the energy level at CitiField for those games was off the charts. Limiting great players performances is bad for the game, but the risk of Harvey getting re-injured is potentially an enormous legal liability in terms of what Boras would sue the Mets for if Harvey is injured; e.g., the cost of lost future employment opportunity ($300 mil) plus emotional and physical damages, which are more difficult to quantify but would represent additional millions.
The Mets trying to toe the balance between unknown risk of re-injury and maximizing Harvey’s contribution to winning baseball games is indeed a tricky business.
If the Mets don’t think the extra innings are dangerous for Harvey, then they should offer him a contract extension–then Matt & Boras might not be so concerned…
I don’t think the Mets would have any problems buying out Harvey’s arb years and even a few of his FA years. The roadblock I think would be Boras and their ability to agree to a number.
Locking up Harvey for the next 4-5 years would also make him a more valuable trade chip over the new few years.
Unless they give him far more than they should, there’s no way Boras and Harvey would take a contract extension.
As far as I’m concerned, the division is locked up. They can go baby Harvey until the playoffs start.
What kind of post is this, with no mention of Harvey’s cowardice and betrayal of his team; Boras’s narrow-minded self-interest; or Alderson’s courage in standing up to them both? C’mon, Dave!
Actually, it’s nice to see some clear thinking on this matter. Thanks, Dave.
I wonder whether the Mets would be better off pitching Harvey and DeGrom in the same game than having guys like Robles pitch in high leverage situations. Harvey for 5 and DeGrom for 4 might be a better way to win games while saving their arms than Harvey and DeGrom pitching on 11 days rest and losing because of a bullpen failure.
It’s hard to argue with these stats, so I won’t try. I’m one of those people who thinks pitchers are being babied these days, I’m willing to be swayed. But I also think you can’t baby pitchers and then suddenly get tough on them for one or two games if being babied is what they’re used to.
I also think that if not for a few unfortunate plays, Robles gets out of that inning allowing only one run, maybe none, and we’re not having this discussion.
But going back to what I said in the first paragraph, I just looked at two pitchers who certainly weren’t babied, as I put it, Juan Marichal and Tom Seaver, and their numbers when facing a batter for the third or fourth time were not much different from the early-in-the-game numbers. Do those stats hold up for most pitchers in the pre-babied era?
This doesn’t sound very “willing to be swayed” to me.
Seriously, Phillies113, is that the best you can offer? I asked a question. Do you have a constructive answer?
I have an answer, but I don’t know if it’s constructive or not. I’ll leave that up to you!
I can’t pretend to know too much regarding the study of the TTO penalty, but a lot of analysts have confirmed the effects time and time again, so I’m willing to put my trust into the numbers and studies. Cherry-picking two examples that buck the trend isn’t enough to convince me that anything is different now, in the so-called Babying Era, as opposed to the Pre-Babying Era. I’m also pretty sure Marichal and Seaver were included in those studies that confirmed TTO.
I agree that teams should handle the starters consistently over a full season if they’re subjecting them to an innings limit, rather than shutting them down towards the season’s end. But your choice of words in your original comment (namely, the overuse of “babying”) suggested to me you’re not as willing to ease up on your stance regarding pitchers’ limits as you said you were.
I don’t have much evidence here, but it looks like it changed in the mid-90s. My guess would be that lineups got a lot deeper at that time, and pitchers had to pitch at full effort and repertoire from the beginning of the game. This is more of a hypothesis than a claim, though.
To further hypothesize without making a claim, it seems like that’s when pitching staffs got deeper as well. Every team started having closers, set up guys, and left/right specialists. As bullpens got better, there was less need to push starters. And as starting rotations got deeper, there was less need to push #1 starters.
In reply to your original answer, in looking up both Seaver and Marichal on B-R, Seaver had an OPS of 613, the 1st time, 621 2nd time, 639 3rd time and 632 the 4th time. Pretty amazing w virtually no decline.
Marichal did have a more traditional “times through the order” penalty. For him, he was 579 the first time, 619 the second, 654 the third and 681 the 4th. All good, but each one a marked decline.
For every Marichal and Seaver there were hundreds of pitchers who weren’t as successful and/or couldn’t remain healthy when pitchers weren’t “babied.” You’re still pointing to small samples when you’re talking about an n of just 2 pitchers. Those 2, for example, never had elbow surgery, as Harvey already has so, if you’re going with an n of 2, why those 2?
There’s a great line in “Bull Durham” where Crash Davis asks, I think of Annie Savoy, something like “why is everyone always someone famous in their past life? Why isn’t anyone Joe Schmo?” The same applies here. When people complain about “the babying of pitchers” everyone always compares pitcher X to one of the All-Time greats. Well, the overwhelming likelihood is that pitcher X is much more comparable to someone who’s not Marichal, Seaver, Ryan, Gibson, or Clemens. It’s much more likely that he’s similar to someone who can’t pitch 300 great innings in a season or someone who has undergone elbow surgery. Your analogy simply isn’t realistic, in all probability.
Max is right. It’s a very tricky situation with no real science to serve as a guide. It’s come up before with Strasburg and will come up again. And not just with TJ cases, with all young pitchers. Dwight Gooden pitched 277 MLB innings as a 20-year old and 276 innings at age 21. When do you think you’ll see that again? The Mets are trying to figure out a responsible way of dealing with the situation and have been innovative in experimenting with a 6-man rotation, skipped starts, shorter outings, etc. As for Harvey, give him a break too. He’s obviously conflicted between his concern for his team, his long-term health, the financial realities of being a pre-free agent star in modern baseball, getting advice from Boras, his doctors, and probably 1,000 different people every day. Maybe 20 years from now there will be an established protocol for these situations but right now there isn’t.
Phillies113: I can’t find out how far back that study goes. I wasn’t cherry-picking. I was bringing up two examples and wondering if the numbers from back then were similar across the board. As for my use of the word “babying,” I’m not sure what other word would be apt. But it does seem to me that if a person is trained to do a certain thing, he might not be successful if he’s asked to do another thing. You wouldn’t train for a half marathon and then try to run a marathon.
In last night’s ESPN broadcast, the announcers pointed out that Harvey’s velocity was down in the fifth inning. It’s odd that that hasn’t been brought up in the convenient columns that people like the guy from Newsday and Vaccaro from the Post have been writing. But why would I expect them to be reasonable? Nobody has brought up that Beltran’s at-bat (in the fourth inning, I think) might have cost Harvey an extra inning.
But I do know that Harvey’s innings limit is not something new; I was hearing 180 to 190 innings as far back as April. I was shocked to see that people were reacting as if they’d never heard it before.
I see. I apologize for being too critical earlier; the term “babying” has more of a negative connotation, so I presumed you were more judgmental than you actually are. As a suggestion, maybe replace “babied” with “handled delicately”?
You have a point regarding what pitchers are and are not used to. It’s one thing to put an innings limit on these guys. But then to have them go full speed for the season, then suddenly stop once they hit the innings limit, seems very counter productive. And who knows? Maybe Harvey would’ve gotten shelled had he been left in. Then people would be yelling that he should’ve come out. There’s no winning sometimes, especially w/r/t the New York media.
No problem. The coverage today in the NY media is driving me crazy; I don’t know why I allow it to. Robles allowed a grand total of one earned run in that inning. The home run shouldn’t have even happened because he should have been out of the inning by then. Of course, I’m sure the Mets have had more than their share of fortunate innings this season, too.
180 to 190 innings had been talked about all the way back in spring training. But according to the Mets, it had always been a flexible thing. Not a hard cap of 180 that Boras started talking about in late August to both the Mets and reporters. That’s what set Alderson off (along with the threatening language Boras had used).
At any rate, Alderson said today that at the last conference call he had with Dr. Andrews and Boras just last week, no hard cap of 180 was ever mentioned by Andrews.
http://www.newsday.com/sports/columnists/anthony-rieber/innings-limit-for-mets-matt-harvey-turns-into-perfect-storm-of-failure-1.10869268
So apparently, Boras lied to Alderson, Jon Heyman and other interviewers.
Wrong URL given above. Here’s the correct one:
http://www.nj.com/mets/index.ssf/2015/09/how_mets_and_matt_harvey_created_his_special_plan.html#incart_river
When I first heard Harvey may have some piggyback starts I thought maybe Dave will finally get the playoff strategy he is advocated a couple of times: starter no more than 2 times through then a reliever parade.
Robles has been pretty good and Harvey obviously is having his innings managed so I wasn’t too down on sitting Harvey after 5. He was pitching on a lot of rest and cruising so I think he likely would have handled another inning fine, but I get it. The error parade that immediately ensued when Robles took the mound guaranteed a LOLmets response from the tabloids etc.
Does time through the order data take into account that there might be a greater % of top of the order hitters in that 3rd time through the lineup sample (as pitchers get pulled sometime during that 3rd time through the lineup)
The 1st time through the lineup – a starter is almost always seeing every batter, including the pitcher in the NL
The 2nd time through the lineup – a starter is probably almost always again seeing the entire lineup, and probably the pitcher again too.
The 3rd time through the lineup – starter may only be seeing the first part of the order (which are presumably better hitters). He may also be seeing PH’s at this point for the weaker hitters at the end of the lineup or at least a PH in the pitcher spot if he makes it that far.
Does the data normalize/correct for this?
Also, is the times through the order penalty related more to fatigue or to the batters having seen more pitches from that pitcher on the day? Major league hitters are the best in the world and I have to imagine they make adjustments throughout the game as they see more and more pitches from the same guy.
Love keeps it together.
For me, the bottom line on this issue isn’t how effective Harvey or his replacements are. It’s really about what expectations are for adults.
I simply refuse to believe that, if innings were an issue, that Harvey did not know about it earlier in the season. If he’s concerned about his health, he should have brought it up at the beginning of the season, or once it became clear that the Mets were contending in July or so – not out of the blue in September after the typical roster moves have been made, and going out of his way to make it a media circus. And remember, he was publicly lobbying to return early last year and implicitly criticizing the Mets for opposing that.
Well as a Yankee’s fan I fully endorse the decision to remove Harvey after 5 innings!
And I agree with Dave that if the goal is to limit Harvey’s workload then lifting him after 77 pitches in a game that (bragging rights aside) was just not that important was the right move.
But – I can’t really get on board with the idea that starters should generally never be allowed to go through a lineup three times or that the third-time through effect “applies to everyone, even pitchers who are throwing really well early on.” Just for kicks I looked up the career numbers for David Price and Clayton Kershaw (to pick two guys at random 🙂 ). Price’s numbers show little variation even through the 4th time through the order. Kershaw’s numbers for the 2nd and 3rd time through the order are about the same and a little worse than the first time through but still really good. And, his numbers for the 4th time through the order are better than for the 2nd and 3rd time and just about as good as for the first time!
Now there’s guys like Price and Kershaw and then there’s mere mortals whom I have every confidence are a lot more subject to the TTTO effect. But Harvey is a lot more like Price and Kershaw than he is like the rest of the field so absent the TJS recovery, importance of the game factors I would leave him in there until he shows signs of fatigue or ineffectiveness.
I dispute the notion that a random middle reliever will always be a better bet than a starter who is on his game facing the same batting order for the third or fourth time. I mean Dave, its not like starting pitchers going through the order more than twice is like breaking the sound barrier back in the pre-Chuck Yeager years. Many multitudes of guys have done this many, many times. You have to judge the individual situation even while being informed by the relevant statistics. Would you pinch hit for your cleanup hitter in a close game after he’s hit 3 homers because the statistics show that only a handful of guys have hit 4 in a game?
Yes, most starters are going to be less effective the third time through the order and even Price and Kershaw will have days where its wiser to pull them after 5 or 6 innings. But less effective doesn’t mean ineffective and limiting the innings that your (typically mediocre) middle relievers throw compared to the guys in your rotation has always been considered a good strategy.
Except the above chase compares their mean performances. Players aren’t their averages.
You have to know that when you take out a rested Harvey at 74 pitches for a brand new reliever that there is a chance he will meltdown. And Robles did. Case closed.
Nothing is proven by knowing that there is a chance of something bad happening and then seeing that bad thing happen. You need to know how likely that bad thing is to happen, how bad that bad thing really is, and how likely something bad was to happen even if you hadn’t made the move that made possible the particular bad thing that you’re talking about.
Was it just me or did Terry Collins hint at retiring at the end of the season last night ?
Interestingly, Stephen Strasburg is an exception to the “times through the order” penalty idea. His career splits:
1st time facing a batter: .226/.284/.357/.640
2nd time facing a batter: .223/.273/.342/.615
3rd time facing a batter: .236/.285/.354/.640
His K/BB actually improves a bit after the first time through. It goes from 4.46 to 4.91 to 4.86.
This season it’s not that he just stays essentially the same – he’s actually gotten better each time through the order:
2015 1st time vs: .238/.297/.393/.690
2015 2nd time vs: .241/.272/.398/.669
2015 3rd time vs: .218/.252/.356/.609
I’m not sure what it means, or what about Strasburg that he holds up so well three times through the order. He’s not immune to the principle: I note that both the career and single season numbers worsen considerably when he faces batters for a fourth time.
Shouldn’t we regress times-through-the-order splits just as we regress other in-season splits? I doubt Harvey’s true talent has such a big jump for the third time through the order. Using in-season splits without regression is a flawed methodology.
A few thoughts; first, if there’s a villain in this story, it’s clearly Scott boras who for some reason felt the need to make this a public debate. The Mets don’t benefit from this. Harvey, his client, doesn’t benefit from this. It’s unclear how/if Boras benefits from this, but he has several other clients who might. At the risk of taking a conspiratorial view… Second, as has been demonstrated in this site, innings is a superficial and largely useless measure for a pitcher’s work, unless the true hardship is crossing the baseline to and from the dugout to the mound. Pitch count, and stressed pitch count is far more relevant. With the kind of money involved, you’d think everyone in a decision making position would have long since come around to this. Third, Hansel Robles has been better than decent. He’s the guy I hoped to see in there if Harvey was taken out. Making a play on defense would have been helpful.
Hey dave, are there guys who are relatively less affected by this third pass through the order effect? if so, do they have anything in common (good heat, depth of repertoire,multiple arm angles etc)?