For the Mets, It’s All About the Rotation

As we move tantalizingly close to pitchers and catchers reporting dates, with many players already descending on spring training complexes in Arizona and Florida, the New York Mets are beginning to draw a little buzz. It’s understandable, but there’s just one problem — those pesky projection systems. While the three systems that I regularly reference don’t bury the Mets, they aren’t exactly pushing them into any sort of pole position — be it for the National League East or Wild Card. That’s not to say the Mets can’t get back to October for the first time since 2006, but much will depend on which version of the starting rotation they get.

Let’s start with where the various systems see the Mets:

FanGraphs – 78-84
Baseball Prospectus – 81-81
Clay Davenport – 77-85

The latter two have the Mets finishing second in the NL East, while here the Marlins are pegged for second and the Mets third. If you peep our Totals page, you’ll see that the Mets are in the bottom 10 teams in both projected WAR for both batters and pitchers. The projections for pitchers might seem particularly low — they’re pegged to have the 20th-best rotation in the game, which doesn’t really befit all of the plaudits their rotation has received. But if we rewind the clock to 2014, we see that this really isn’t too far removed. Last year, their starting rotation ranked 25th in WAR. They essentially had three two-win or better starters, but they also had three sub-replacement level starters to drag their total down, including fifth starter Dillon Gee, who they couldn’t give away this winter.

The pick of the litter was Jacob deGrom, who tallied 3.0 WAR in roughly three-fourths of a season. If he can just keep that up the Mets have a superstar on their hands, yes? Sure, that sounds great in theory, but it wouldn’t be very wise to set that as the baseline. Even the oft-optimistic Fans projections peg deGrom at the same 3.0 WAR despite adding more than 50 innings pitched to his projection. ZiPS is in the same neighborhood. Steamer and PECOTA though are much more pessimistic, with both clocking in around 1 WAR(P).

In fact, the Steamer and ZiPS projection systems paint two distinct pictures of the Mets season. Steamer paints them as having one legit good pitcher, and a bunch of guys who can be good on any given day but at the end of the season won’t amount to much. It basically takes last year’s rotation, and replaces ace deGrom with ace Harvey. ZiPS, on the other hand, paints them as having two leading men in Harvey and deGrom and a strong number three in Zack Wheeler.

Certainly, the Mets’ prospects have the talent to make Mssr. Szymborski look incredibly right. Guys like Noah Syndergaard, Rafael Montero and Steven Matz have been top prospects for a reason. There are two questions though, with respect to 2015. The first is how quickly are they going to get a shot. After two straight seasons with a FIP- that is at least 10 percent worse than league average, we can safely paint Gee as a guy who doesn’t really need innings. Bartolo Colon has been gilded as a lovable personality in recent years, but he was league average last year, and at age 42 it’s an open question as to whether he’s the right person to be taking innings in the rotation. In other words, if the best Gee or Colon can be is league average, why not cut bait and give run to Syndergaard or one of the other young bucks?

That leads us into the second question — can the younglings be good right away? Plentiful are the number of starting pitchers who weren’t exactly lights out in their first season. We need to look no further than new Hall of Famers Randy Johnson, Pedro Martinez and John Smoltz, who all rocked walk rates of over 10 percent in their first real seasons of big league ball. Success is more likely for these Mets prospect pitchers than most, but the timing is an open question. Mets manager Terry Collins has a juggling act ahead of him.

If ZiPS is more right about the Mets rotation, and Collins plugs in the right pitchers into the rotation, the Mets will indeed have one of the top rotations in the game. Which is good, because if they want to contend, they’re going to need it. The team has more obstacles to overcome than just some incongruous pitching projections.

Defense
Teams have to play defense, and unfortunately the Mets don’t project to play it very well. The only starters who project to be above average defensively are Juan Lagares and David Wright, and Travis d’Arnaud was among the league leaders in pitch framing last year, which is also good. Curtis Granderson and Lucas Duda shouldn’t hurt the defensive effort too much, but they aren’t really going to help either. Michael Cuddyer, on the other hand, is going to be a significant detriment. ZiPS has him projected for a -8 Def in just a half season of action (372 plate appearances). Steamer has him at -8.4 in more action (591 PA, or 138 games). That’s slightly more encouraging, but the bottom line is that Cuddyer is going to hurt the team defensively, and hurt them badly.

That leaves the up-the-middle infield defense, which looks to be the real sinkhole. The projection systems agree that Daniel Murphy and Wilmer Flores will be very bad, and that Murphy will be a bigger problem than will Flores. But both will be a problem — ZiPS sees Flores as a -7 defensively, while Steamer is at -1 and the Fans are at -3. Using the ZiPS projections as a guide, we actually find that the Flores/Murphy combo is projected to be the worst in the game. I’ll spare you the whole table and just give you the top five and bottom five:

  • Atlanta: +19 (Simmons/Gosselin)
  • Colorado: +14 (Tulowitzki, LeMahieu)
  • Detroit: +14 (Iglesias, Kinsler)
  • Baltimore: +13 (Hardy, Schoop)
  • Cincinnati: +13 (Cozart/Phillips)
  • Anaheim: -8 (Aybar/Rutledge)
  • Toronto: -9 (Reyes/Izturis)
  • Chi Cubs: -11 (Castro/Baez)
  • Houston: -11 (Lowrie/Altuve)
  • NY Mets: -15 (Murphy/Flores)

As you can see, the Mets are just one of three teams to be projected in the negative double digits, and they have some pretty decent separation from the rest of the bottom of the pack as well. Not exactly a pretty picture. It should be mitigated slightly by the fact that only Zack Wheeler and Jon Niese generate a high number of ground balls, but plenty of balls are going to find Flores and Murphy, and it’ll put a strain on the pitchers when they don’t get to them. Dilson Herrera and Matt Reynolds are looming, but Kiley sees Herrera as only average defensively at second, and Reynolds as more of a second baseman than a shortstop. The Mets might not have been able to do better than Flores, but combine him with Murphy and it isn’t pretty.

Offense
According to Steamer, here are the number of Mets hitters projected to have a .340 wOBA or better: One. David Wright. According to ZiPS, here are the number of Mets hitters projected to have a .340 wOBA or better: One. Lucas Duda.

The Fans projections are a bit more optimistic — they have Wright, Duda, Cuddyer and d’Arnaud at .340 or better. Complicating this even further is that while Duda should be fine, Cuddyer, d’Arnaud and Wright have spent a good amount of time in the trainers’ room in recent seasons. Beyond them, the team is short on guys who you would expect to smash their projections. If Cuddyer, d’Arnaud and Wright can make it through the whole season intact, the Mets should have a reasonably acceptable offense. If they can’t, the Mets might have trouble scoring runs, no matter how far the fences come in.

Bullpen
It was a nice story that the Mets finally found a role for Jennry Mejia last season, especially after all the dicking around the Mets did with him in the past. But Mejia isn’t exactly a top notch reliever, at least not yet. And along with Bobby Parnell and Jeurys Familia, he is one of three expected Mets relievers who are projected with a positive WAR. And we really don’t know what we’re going to get from Parnell. He is well removed from surgery, so he should be back to normal, but there is still lingering uncertainty there. And it’s not like he was Eric Gagne in his prime or anything before he got hurt. Realistically, the Mets aren’t going to make a lot of noise with this bullpen. Upgrades will be needed, but who knows if they’ll have enough “payroll flexibility” to make them.

The Mets didn’t have a very good rotation last season, and they might not this year. There is a wide berth between what they are capable of doing and what they are likely to do. But given the team’s problems elsewhere — a leaky middle infield, an offense led by players with ugly injury histories, a middling bullpen and their owner’s current financial state — their playoff hopes are going to live and die with the starting rotation.





Paul Swydan used to be the managing editor of The Hardball Times, a writer and editor for FanGraphs and a writer for Boston.com and The Boston Globe. Now, he owns The Silver Unicorn Bookstore, an independent bookstore in Acton, Mass. Follow him on Twitter @Swydan. Follow the store @SilUnicornActon.

57 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Noah Baron
9 years ago

The reason the Mets rotation is projected to be poor in 2015 has little to do with the Mets rotation and everything to do with the flawed way that FanGraphs calculates Pitcher fWAR.

First, FanGraphs uses a runs park factor when calculating pitcher fWAR. That would be fine if WAR was based off of RA/9, but it’s not. Citi Field is actually a neutral park when it comes to FIP. This mistake significantly underrates the entire Mets pitching staff, both starters and relievers.

Second, the way FanGraphs calculates replacement level FIP is also incorrect, and causes National League pitchers to be underrated. Rather than using the NL league average FIP when calculating WAR, FanGraphs uses the league average ERA and divides by .92. While this might seem like a minor issue, the NL almost always has a FIP higher than their ERA with the AL experiencing the opposite phenomenon.

I’ve actually written about both issues on Community Graphs, and was promised that the first issue would be fixed by using FIP park factors instead. So far, nothing has changed, and I’m losing faith that anything will change.

The only good side to this whole fiasco is that websites with poor analytical skills (who haven’t yet discovered the fountain of problems that riddle fWAR) cause the Vegas World Series and win total odds to be way off, allowing some of us to make a lot of money.

It still amazes me how nobody besides me realizes there is a problem when Mark Buehrle is projected for more WAR than Zack Wheeler despite having a FIP .81 runs higher. The difference between the league average AL and NL FIPs last year was only .10 runs.

It’s just odd to me how far FanGraphs has deviated from their original goal of outside the box thinking and statistical analysis. Now it’s complete reliance on a flawed projection system (Steamer) and an inability to improve from within.

David Appelman
9 years ago
Reply to  Noah Baron

I am late to this thread, but I wanted to address some concerns about us not changing our FIP park factors.

First off, I really appreciate Noah’s work in pointing out this particular quirk of park factors. We’ve had fWAR up since 2008 and this is something which was never addressed, at least as far as I remember, so kudos to him for pointing it out.

I personally did not get a chance to dig into the suggestion until late December, though I was aware that it was published because I specifically requested it be published once I saw it hit the community research queue (we don’t publish everything). I wanted some community feedback on his suggestion. Noah’s post was published on December 12th.

While Noah’s proposed posts sentiment was accurate, the methodology for calculating park factors he suggested needed some looking into. After some digging internally, we decided that the proposed park factor calculation had some issues and we decided to calculate FIP park factors from scratch. These were published here on December 30th. 2014 FIP Park Factors Revised

In terms of how we update stats on the site, generally we can do it pretty quickly, sometimes even the same day. However, WAR is a somewhat different beast of a stat. WAR is a complicated stat and changing it takes more work than every other stat on the site. Also, since WAR is such an important stat to us and the baseball community, when we make changes we try to do so in a deliberate manner.

We’ve had plans to make some changes to WAR this off-season and there were a few other changes we had planned on incorporating into the stat, not just the park factor changes. These have been taking a little longer than expected.

To briefly address whether or not any of this impacts our team projections, the park factor issue really only impacts WAR (and FIP-). Our team projections, in terms of projected record do not rely on WAR at all.

We hope to have the park factor changes and a few others rolled out within the next week. Thanks for your patience!

(This was posted on Feb 19th, 2015 @ 12:16AM)

huric9
9 years ago
Reply to  Noah Baron

Noah, I agree completely!

ChummyZ
9 years ago
Reply to  Noah Baron

Was expecting this in the comments section. Was not disappointed. You’re 100% correct, and your articles were great. This really needs to be fixed ASAP because of this site’s extreme reliance on WAR.

Slightlyawkward
9 years ago
Reply to  Noah Baron

My problem with your whole fWAR park factors crusade is that the lists of stadiums helped and hurt by this both included hitters and pitchers parks. For example, Citi Field and Yankee Stadium were both considered hurt by this phenomenon. Also, the 2013 Yankees staff was ranked 8th in fWAR and their rotation was ranked 4th. They weren’t that good. It would sound like if this is the reason fWAR is off one year Yankee Stadium helped the Yankees and the next it hurt.

melon130
9 years ago

One of the comments has corrected FIP park factors, looks like Yankee Stadium isn’t as badly affected as was originally stated.

So what you're saying is
9 years ago
Reply to  Noah Baron

That Citi is a neutral park? I suppose you also think Las Vegas is neutral as well…

ChummyZ
9 years ago

Any semblance of reading ability would show that he said “by FIP.” I guess you don’t possess that.

2014 Citi Field Park Factors for basic FIP inputs (TTO):

HR: 102
SO: 101
BB: 100

Scoko
9 years ago
Reply to  ChummyZ

What about prior years? Where can I see park factors based on FIP?

Noah Baron
9 years ago
Reply to  ChummyZ

While FIP park factors aren’t currently available on FanGraphs, all you have to do is look at FIP’s component park factors – the ballpark’s HR, SO, and BB park factor values. Since the fences were moved in 2012 Citi Field has been a neutral to slightly hitter friendly park factor with regards to FIP.

ScoKo
9 years ago

Got it. Not sure if this is going to be a dumb comment im an amateur, but does this even alter overall team evaluations? I see how it does for pitching specifically but what about overall? Doesn’t this just mean the pitching is really better than their WAR indicates and then on flip side wouldn’t it mean their offensive WAR is really worse, which would negate in an overall evaluation?

Noah Baron
9 years ago
Reply to  ScoKo

No, the FIP park factor offense doesn’t affect hitters. For offense you still use a runs park factor. The FIP park factor is only used on the FIP-based WAR; offensive fWAR is wOBA based, so a runs park factor makes more sense.

Essentially, runs park factors should be used for position player fWAR, and FIP park factors should be used for pitcher fWAR. And this has a really major effect on WAR calculation; I recalculated Zack Wheeler’s WAR with the new park factor and got a WAR value almost 50% higher than what is currently being displayed.

Noah Baron
9 years ago
Reply to  ScoKo

Basically, what this FIP/runs park factor difference tells us is that, with a neutral defense, we can expect the Mets pitching as a whole to have a higher FIP than their ERA because of the way FIP is calculated. Ditto with the Pirates, Rays, and Padres. We can also expect the Red Sox, Marlins, Rockies, and Rangers pitching staffs to have higher ERAs than their FIPs.

Yet, for some reason, FanGraphs chooses to go into the 2015 season with bad data. Am I being a bit snarky? Sure. But I fear that if I don’t make a ruckus these changes will never be implemented.

ScoKo
9 years ago
Reply to  ScoKo

Noah, one thing I dont understand is how a ballpark has any relevance on Ks and BB? I am confused how park factors based on FIP make any sense if 2 of the 3 categories stats effect a pitchers fip have 0 to do with ballparks? Dont ballparks effect the things a pitcher cant control, like hits?

ScoKo
9 years ago
Reply to  ScoKo

How can a ballpark rate neutral, pitcher, or hitter friendly in terms of K. I dont get it? Isnt the odds of a K the same regardless of where you are pitching? How has nobody made this point? Saying citi is a neutral park for K and BB seems like a dumb comment to me. Isnt every single park neutral for K and BB?

The Tone
9 years ago
Reply to  Paul Swydan

One would think that improving your methods for interpreting data in order to produce a more reliable projection system would be high up on your list of priorities heading into the season. I have no opinion on how any of these changes would improve or diminish the projections for the Mets’ pitchers or any other team’s pitchers, but it’s kind of flat out un-professional and, frankly, snobbish of you to respond to a reader like that (I refer to the general tone of your response rather than the data you present). I know it’s your article and that you definitely put a lot of work into writing and publishing it (as all FG writers do–which I believe we all appreciate), but you should probably work on your ability to roll with the feedback you get in a more even-keeled way, especially when there is some real substantive value to what the commenter has to say. It’s kind of just the nature of the beast with this line of work. Good luck and keep up the hard work.

Noah Baron
9 years ago
Reply to  Paul Swydan

I understand that you guys have things to do, but it seems to me that fixing WAR should be the first thing on the agenda. Having an outdated WAR system leads to inaccurate team projections.

It’s actually quite surprising how neatly the teams that stand out as being underrated by FanGraphs WAR (Mets, Pirates, Rays) also happen to be teams where this park factor miscalculation exists. Likewise, the Red Sox, Marlins, Rockies, and Rangers all stand out as teams being overrated by fWAR, and, amazingly, they all are being affected (positively) by this miscalculation.

Noah Baron
9 years ago
Reply to  Paul Swydan

i’m sorry if my comment was filled with anger; I don’t mean it personally. Really, this issue has nothing to do with any of the writers specifically (I actually enjoy your writing very much).

I’m a bit of a perfectionist, and seeing a glaring error make the Mets pitching staff seem like the worst in baseball (when pretty much everyone knows it’s not) really aggravates me.

Noah Baron
9 years ago
Reply to  Noah Baron

I assure you that fantasy is interesting to me; I run an otteneu team, and otteneu certainly warrants an update.

BurleighGrimes
9 years ago
Reply to  Paul Swydan

It seems fair that fangraphs has other projects or priorities in order than fixing how pitcher value is calculated and projected, but I believe it is intellectually disingenuous to use a projection system that you all have admitted has flawed inputs and needs to be fixed. If you want to use the projections, you have to fix them right away. If other things take precedence, there has to be a moratorium on using numbers you know to be flawed.

Because I have followed Noah’s research, I know about this systematic flaw; it makes it hard for me to take fangraphs pitcher WAR or projections at face value, which is frustrating.

Hank
9 years ago
Reply to  Noah Baron

This is a really good point – and one which I have asked several writers here without getting a real good response.

Miami is a similar issue (nearly run neutral, massive HR suppression); if you look at their starting rotation FIP- and xFIP- over the last 3 years there appears to be a consistent offset. While there will be some variation expected, these should start to look similar at some point (an entire rotation for 3 years seems like an adequate sample) Other stadiums with disparate HR and run factors also can have this issue (YS, Fenway, maybe Cincy?)

I don’t think this is as much an issue with projections though. I believe it’s more an issue with fWAR (the FIP based one obviously) and FIP-. The projection systems address this better with better HR projections – though converting it into projected WAR can still be problematic.

Noah Baron
9 years ago
Reply to  Hank

That’s really the issue. While I disagree with some of the projections, the ZIPS projections specifically are very reasonable.

The problem is when the FIPs are then converted into projected WAR. It’s why Buehrle can be projected for almost 50% more WAR than Zack Wheeler despite a FIP that’s 0.81 runs higher.

tradeharrychiti
9 years ago
Reply to  Noah Baron

Thank you for sharing this!

The Mets bullpen and rotation is loaded with talented young pitchers (and, as noted in the article, there are a few more high-upside arms waiting in the wings).

I think the projections for the Mets pitching staff referenced in this article are particularly pessimistic. Their staff could be a top three in the NL this season.

Jared Crossmember
9 years ago
Reply to  Noah Baron

Noah, I’m with you on the FIP park factors issue and would like to see that change.

The reason, however, that people use flawed projection systems or flawed models more generally is, of course, that perfect models don’t actually exist and we’re often better off with a flawed model than with no model at all. For instance, in this case, Steamer projections, while imperfect, are considerably more accurate than your impressions.

CrazyPants
9 years ago
Reply to  Jared Cross

Beg to differ. People will stop referring to models that give highly questionable results, and go back to using their own projections like we did in the dark ages of a few yrs ago.

Jared Crossmember
9 years ago
Reply to  CrazyPants

And should if their own projections are better. In this case, they aren’t.

Noah Baron
9 years ago
Reply to  CrazyPants

Not all of us have the time to make our own projection system. That doesn’t mean we can’t look at what a projection spits out and realize that the projection isn’t very useful.

And really, I’m actually a fan of Steamer projections. For the most part it’s better than people’s subjective opinions. But it also doesn’t utilize every piece of information, so let’s not treat it as an authority (which seems to be the trend here at FanGraphs). It’s a work-in-progress with perfections and imperfections, even if it does treat all players uniformly.

Jared Cross
9 years ago
Reply to  CrazyPants

I don’t think utilizing every piece of information or even all that much information is required for an algorithm to be better than my subjective opinion. Marcel uses very little and yet it throttles subjective opinion when it comes to accuracy. Left with my opinion or Marcel I should reject my opinion and take Marcel every day of the week. Now, in this case we have Steamer and ZiPS and ZiPS is more optimistic so there is reason to think that Steamer is low for the Mets but my claim is just this – when left to pick between my subjective opinion and any reasonable algorithm, I’d be better off with the algorithm. Marcel does better NOT using additional information than my subjective brain does when using it.

Noah Baron
9 years ago
Reply to  CrazyPants

I think you’re looking at this the wrong way. I’m 100% with you on using projection systems over subjective opinions. But it doesn’t have to be that cut and dry; most of the “subjective opinions” you’re referring to aren’t just subjective, they’re ill-informed, relying on useless data (like ERA, RBIs, Wins, etc.) to form their opinions. Even if a computer projection system that has the ability to beat these subjective opinions (like ESPN, for example), the system itself might not actually be able to beat extremely well-informed subjective viewpoints.

Clay Davenport, who to my knowledge doesn’t have a computer projection system, had more accurate projections than FanGraphs (which uses Steamer) last year. The problem might not actually be Steamer, but instead the conversion process from Steamer into WAR.

Either way, it’s hard to put a ton of faith into a projection system that a) doesn’t reveal its methods and b) doesn’t always outperform human projections.

Noah Baron
9 years ago
Reply to  CrazyPants

But, if you’d like, I’ll release my own projections as a way to see if Steamer projections really can be beat. Do you accept my challenge?