Diamondbacks Take a Two-Year Chance on ERA
Kevin Towers is a confident, talkative man. He’ll tell you what he’s thinking about, and he’ll tell you what he’s doing. There aren’t many anonymous leaks that come out of the Diamondbacks organization because Towers doesn’t exactly keep many secrets — even when he’s actively negotiating. Towers has spent much of this offseason talking about how badly he wanted to acquire a No. 1-type starting pitcher. He was in on Masahiro Tanaka; he’s been in on David Price. He admitted he wanted to make a significant splash. Just the other day Arizona signed Bronson Arroyo for two years and $23.5 million. Or three years and $30 million. The bigger point is that Arizona signed Bronson Arroyo, and now they’re finished.
Arroyo, of course, is no one’s idea of an ace. Most simply, the best pitchers get strikeouts, limit walks and limit dingers. Arroyo does one of those things. He’s 37 in a couple of weeks, meaning he’s basically 37 now. There’s little sexiness with this acquisition, and Towers would tell you he knows he didn’t get a No. 1. Still, Arroyo does have something going in his favor. It’s just a matter of how much you believe in it.
Well, I suppose Arroyo has two things going in his favor. One is that, so far, he’s been extraordinarily durable. In nine straight years, he’s made at least 32 starts. He’s drawn not-inaccurate comparisons to Livan Hernandez. His injury history is enough to make you chuckle: There’s a bit of back soreness, there’s a bit of leg soreness, there’s carpal tunnel, there’s mono and there are four contusions from batted balls. Arroyo’s been the image of dependability, and people credit his work ethic and his smooth, controlled pitching style.
Now, that’s meaningful. But it’s only so meaningful. Hernandez never got hurt. Dan Haren has begun to have a few problems. Jon Garland went straight from reliable to shoulder surgery. You can even point to Roy Halladay. There’s no such thing as an automatic 200 innings. Absolutely, Arroyo has a lower injury risk than Erik Bedard. But he still has some kind of injury risk because he’s still throwing 3,000 pitches a season, and then some.
The other thing Arroyo has going for him are his actual results. The past five seasons, 141 starters have thrown at least 400 innings. Right there, you know this pool is selective for starters decent enough to stick around. In WAR/200 innings, Arroyo ranks in the bottom 10, at 0.9. He’s in the company of guys like Barry Zito and Nick Blackburn and Kevin Correia. In RA9-WAR/200 innings, however, Arroyo ranks 68th — middle of the pack — at 2.6. The difference between those two numbers is 1.7, and that difference is third-greatest over the span, behind Clay Buchholz and Jeremy Hellickson. In short, Arroyo’s been beating his peripherals.
On our leaderboards, FDP-Wins is the difference between RA9-WAR and WAR. Between 2009 and 2013, Arroyo led baseball in FDP-Wins, and he led by almost two over Jered Weaver. Arroyo even beat his own peripherals in his disastrous 2011 campaign. That year, he was worth -0.3 RA9-wins, but he was worth -1.5 FIP-wins. If you look only at Arroyo’s walks, strikeouts and dingers, you see a guy who might be worth a few million on a one-year contract. If you look at the rest, you see a guy who’s better than that.
This is always such a nervous, uncomfortable thing because we never know quite how much stock to put in a guy’s ability to beat his FIP. Arroyo routinely posts better-than-average BABIP marks. He also pitches around the edges, and this implies worse-than-average quality of contact. But then, there are all of the home runs, which Arroyo allows, and which he will forever allow. It’s easy to see how Arroyo could be more deceptive than most, but again, the homers, and the overall contact rate. The true Bronson Arroyo almost has to be somewhere in the middle.
Between 2008 and 2012, the big-league leaders in FDP-Wins were Tim Hudson, Jeremy Hellickson, Matt Cain, Bronson Arroyo and Jered Weaver. In 2013, Hudson undershot his peripherals. Hellickson did, too, by quite a lot. Cain was right around even. Arroyo was Arroyo. Weaver posted an FDP of 0.9. There’s some ability here, but it’s hard to separate from all that noise, and this is why the Diamondbacks are taking a gamble. The contract comes with limited upside, and while it’s not bad, there are more ways for it to go wrong than right.
Towers did well in addressing the need for depth. He recognized the Diamondbacks would probably need more than five starters, especially given their rotation includes Brandon McCarthy, and that Archie Bradley is the only helpful guy on the horizon. Granted, this was an issue partly of Towers’ own making. But while there might not be an enormous difference between Randall Delgado and Arroyo, Arizona’s probably going to need them both. Delgado will open out of the bullpen, but he’ll start some games unless the team gets a whole lot of good luck.
Arroyo probably makes the Diamondbacks better. It depends on how much you believe in his runs-results. And while pitching in Arizona won’t do him many favors, given his style, he is coming from Cincinnati. The contract, also, isn’t potentially devastating in any way, and you can at least count on Arroyo to pound the zone and get plenty of called strikes. Yet what one has to wonder is whether Towers could’ve spent better than this. Arroyo got two years, guaranteed, at $23.5 million. Paul Maholm got a guaranteed one year for $1.5 million, plus incentives. Maholm knows he might end up pitching out of the bullpen. The past three years, Maholm has Arroyo beat easily in WAR, and they’re even in RA9-WAR. Jason Hammel wound up with a cheap one-year contract. Dan Haren got half of Arroyo’s contract. Scott Baker got a minor-league contract. Shaun Marcum got a minor-league contract. Arroyo got Tim Hudson money, but Hudson seems like the safer gamble.
Arroyo’s deal looks like a contract with significant downside. Arroyo seems like he could come apart at any moment. He seems like he’s pitching with just about the narrowest margin of error. He’s also allowed 174 runs in his last 404 innings, pitching half the time in a hitter-friendly ballpark with utterly underwhelming stuff. In the end, this is a two-year bet on continuity of a very particular and unreliable kind.
Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.
I wonder, some of Arroyo’s best pitched games last year were against NL West teams. Weak hitting in general in the NL West, but maybe Towers factored that into the decision?
Looking at the 8 non Reds/DBacks teams in the 2 divisions, we find these 2013 rankings by wRC+
7. Cardinals
8. Dodgers
11. Giants
13. Pirates
17. Padres
18. Brewers
22. Rockies
26. Cubs
The other 4 NL West teams combined for basically league average wRC+, and they were better than the 4 Central teams overall.
I don’t know where you are getting your narrative about the NL West having particularly weak hitting vis-a-vis the NL Central.
Park adjustments are an amazing thing … Rockies scored second most runs in the NL last year (very close to first), and yet are 22nd in baseball in wRC+. And who would have guessed the Giants would be top half in baseball offensively.
Arroyo may be operating on a thin margin, but there is a statistical basis for his success. 1. He doesn’t walk anyone. 2. His flyball tendency keeps his BABIP low. 3. #1 + #2 keep his WHIP at minuscule levels. 4. The low WHIP prevents those 3 run dingers Earl Weaver used to love. Arroyo is able to limit the damage.
I was really hoping Arroyo would sign with the Giants so we could see how pitching in a park tailor made for him would work out. 23 out of his 32 dingers were allowed to LH batters. AT&T crushes LH power!
Good call. Would have enjoyed seeing that as well. Just going from one band box to a park that plays like another.
As a Giants fan, I’m glad they signed Hudson instead, especially after reading this article.
The park does not have any way of helping extreme GB pitchers. I’m OK with Huddy, but I would have been OK with Arroyo too. Even better would have been signing both instead of re-upping Vogelsong who I don’t have a lot of confidence in right now.
Your above points do not explain why he should outperform his peripherals. Peripheral stats like FIP will take into account the fact that allowing fewer walks will mean fewer batters on base if he gives up a home run.
Walks ares only half of the equation. FIP does not account for the low BABIPs being sustainable due to the flyball tendency.
FIP is weird — it’s a linear combination of per-out walks, home runs, and strikeouts, though none of these rates are exactly linearly related to run-allowance. Close to league-average rates FIP is still pretty close to a more complex formula taking into account their interaction; on the extremes FIP is farther away, and Arroyo walks and strikes out far fewer batters than normal…
this looks like one of the poorer FA contracts handed out this winter. there was no reason to get into a bidding war over Arroyo which may be the only excuse AZ had to go as high for him as they did, but that’s just speculation and not an excuse for them to have signed the deal. They were looking to throw money at someone this winter and apparently didn’t want to give up a pick to sign either Ubaldo or Santana. They paid a lot just for durability. Isn’t Oswalt available?
Poor contract? Bidding war? Its a 2 year deal for 11m a year for a guy who has put up a consistent 2.5-3 RA9-WAR for years.
It’s only a bad contract if you think that the guy outperforming his FIP for a decade is just luck.
if they wanted a TOR SP then they should have been willing to pay for it (prob not much more than what they did for Arroyo). Arroyo’s innings-eating ability is one thing but 37 yo SPs who rely on deception and continuing to outperform FIPs are fighting a losing battle. Given the SP pipeline in their org, they should have signed Arroyo for 1 yr, not 2 or god forbid 3.
“it’s only a bad contract if you think that the guy outperforming his FIP for a decade is just luck.”
See Jeff’s comments below. That’s been a 5-year thing, not a decade thing.
Also, there’s about a 6% chance that a player will either underperform or outperform his FIP for five consecutive years just by random chance. Which certainly doesn’t mean that it’s totally random and the pitcher has no control over it; it just doesn’t mean that it’s *not* random, either.
Lol. Did you use a binomial distribution to conclude that and make an absurd conclusion about his results??
So you’re saying that there’s a 94% chance that it’s due to skill?
No. The math is pretty simple. If a pitcher has average BABIP and LOB skills, then there’s about a 50% chance he’ll have a better ERA than FIP due to luck. Multiply that by 4 and you have the chances of that happening for four straight seasons.
.5^4 = .0625
Meaning that if Arroyo has no actual skill to outperform his FIP, there was a 6.3% chance he’d repeatedly outperform it anyway.
With a chance of 6.25%, that also means that with a sample of only 16 pitchers over 4 seasons, most likely one of them will outperform his FIP the whole time based on luck alone. If you are talking about the hundred or so pitchers who qualify over the same time frame as Arroyo, we’d expect 6 or more to outperform their peripherals due to luck alone.
Arroyo has overperformed significantly. There isn’t a 50% chance of that.
If FIP is a good stat (which it isn’t), than I’d guess you’d have less than a 10% chance of randomly improving .5R/9 randomly over the course of a whole season.
“There isn’t a 50% chance of that.”
Good thing no one claimed that then!
This is horrible. Stop with this amateur math hour.
Dude does a killer cover of Pearl Jam’s, “Daughter”…….especially with Peter Gammons playing Rhythm.
This contract may or may not work, but this devotion to FIP and FIP alone needs to be tempered. It is certainly true that “on average” FIP is better at predicting the future than is ERA. However, when someone consistently out-performs their peripherals you should not be relying on averages. Guys do not get lucky over several hundred innings, there is a real skill set there even if we can’t precisely measure it (see Glavine, Tom). Arroyo is not exciting, but he has been consistent. Odds are he’ll continue to be that over the life of this contract.
Last five years, +8.6 FDP-Wins. Career, previous to that: -3.6 FDP-Wins.
His pitching style has clearly changed in the past 5 years, though. He’s significantly decreased his Ks, BBs and BABiP.
From 2004-2008, (his first 5 years as a full-time starting pitcher) he had a 6.3 K/9, a 2.6 BB/9, a 1.1 HR/9 and a .294 BABiP.
From 2009-2013, he has put up 5.3 K/9, 2.1 BB/9, a 1.4 HR/9 and a .269 BABiP.
Those are two different pitchers.
The point is just let’s not be so certain about a guy’s ability to beat his peripherals. Everybody’s always looking for exceptions, which can lead to false positives. The default assumption should always be that a given pitcher is normal. Arroyo might not be, but we don’t actually *know* that, and we don’t know the true magnitude.
” The default assumption should always be that a given pitcher is normal.”
The default assumption, when the results show otherwise for 5+ years, should be that the default assumption may not apply.
Oh, the default assumption is always that the pitcher in question is normal. After you make that default assumption, we then have the ability look into the ten years of data we have available.
I would be interested to see how THT’s fly ball research applies to Arroyo, personally.
Correct, may not apply, as discussed in the article. It’s a question of how much is Arroyo and how much is not.
“The default assumption, when the results show otherwise for 5+ years, should be that the default assumption may not apply.”
False. The default assumption for Arroyo or ANYONE is that the normal expectations apply. 5+ years of outperforming peripherals is useful evidence to help disprove the default assumption, although by no means ironclad.
It is best to remember the old adage: “all models are wrong, but some are useful”. We know FIP is wrong because there is no such thing as an average pitcher. That doesn’t mean that it isn’t useful though.
As far as I can tell, people err on both sides of the FIP debate relatively equally (under estimating guys that beat their peripherals and over estimating guys that have better peripherals than performance). It seems that people are just far more comfortable with particular kinds of being wrong: overestimating a bad performance (i.e. thinking a guy with good peripherals is better than he is) and underestimating a good performance (i.e. thinking a guy that beat his peripherals was lucky). In short, people are always going to bet on the peripherals because it is safer to do so.
“False. The default assumption for Arroyo or ANYONE is that the normal expectations apply. 5+ years of outperforming peripherals is useful evidence to help disprove the default assumption, although by no means ironclad.”
You say false, and then you agree with me.
5 years of significantly overperforming a statistic means that the statistic is more likely to be wrong than the player to actually be overperforming.
FIP is also significantly less predictive than ERA in certain types of parks (mostly hitters parks).
It’s a shitty statistic. Its not fielding independant. It ignores a good percentage of what pitchers can control, and it places too much emphasis on something that pitchers don’t totally control.
Literally the only inputs for FIP are home runs, strike outs, and walks. What is the factor that they don’t totally control that has too much emphasis? Your only option is home runs; xFIP to the rescue! What exactly does it ignore that pitchers can control?
It ignores batted ball data, which pitchers do have an effect on.
Lolololo
But ERA, as has been studied repeatedly, gives pitchers too much credit/blame for BABIP variation, to the point that FIP is a better representation of a pitcher’s skill, despite taking the opposite extreme.
5 years of ERA correlates better to future ERA than 5 years of FIP do.
“Literally the only inputs for FIP are home runs, strike outs, and walks”
No, literally the only inputs for FIP are home runs, strike outs, walks, and INNINGS PITCHED.
Innings pitched is outs/3. Roughly 80% of those outs are affected by defense. In roughly half of pitchers, getting an out via a groundout actually raises FIP.
No, FIP is not less predictive than ERA in those situations, it may just require adjustment of its constants in some extreme situations.
Ballpark doesn’t suddenly make a stat that holds a pitcher responsible for everything his defense does more predictive than one that doesn’t.
FIP is worse at predicting future performance for pitchers that pitch the majority of their innings in hitters parks than ERA is.
This is not arguable. Your lack of understanding does not change this.
Yes they most certainly do.
1. Soft tossers age better than high FBv guys, alleviating some of the age worries with Arroyo. True or False?
2. He’s not beating the variance we might expect between FIP-ERA (0.5) by *that* much. Plus, there is an explainable element to this–as DrB points out–miniscule WHIP on account of the >2.5 K:BB ratio. Not sure about the FB tendencies in ‘Zona, but I guess we’ll see…
The fact that he isn’t beating that variance by much would be significant in a one year sample, but we have a 5 year one.
And he should actually be helped by going to ‘Zona. Cincy is much, much, much worse for home runs, and the other NL Central ballparks are much less pitcher friendly than the NL West ones are.
simple case of diminishing returns should dominate the forecast for Arroyo given his age, place on the velocity spectrum and contact rate he allows. it’s not hard to forecast that this is about where he should begin to fall apart. these guys don’t go on pitching like this indefinitely.
Cin: 112 HR factor, 2nd overall.
Ari: 103 HR factor, 12th overall.
Definitely worse in Cin, but it’s not opposite ends of the spectrum.
Low OBP allowed (basically what WHIP tries to measure) does not explain ERA-FIP discrepancy. Only an unusual distribution of his baserunners allowed explains it. So, if he becomes unusually homer-averse with men on base, and unusually homer-prone with the bases empty, that might explain some some of the discrepancy. However, that is not necessarily a result of skill.
If the low OBP is due in significant part to a sustainably low BABIP due to flyball tendency, then it DOES explain the discrepancy between ERA and FIP.
I dont really know much about Cincinati’s outfield in recent years, but could above average defense contributed to the difference between RA and FIP?
Shin-Soo Choo played CF last year.
I think that says enough about their OF defense.
Here are the Reds’ ranks (out of the 30 MLB teams) in defensive efficiency over the last five years.
2009: 4th
2010: 9th
2011: 3rd
2012: 12th
2013: 1st
Yes, even with Shin-Soo Choo playing center field, the Reds had the most efficient defense in baseball last year.
Here are their ranks in team UZR over the same span:
2009 – 4th
2010 – 4th
2011 – 5th
2012 – 7th
2013 – 8th
So, it looks like Arroyo has emerged as a pitcher who beats his peripherals in the same span that the Reds have consistently played very good defense behind him. The two probably aren’t unrelated.
In 2013 Cincinnati finished 8th in both UZR/150 and DRS, which is pretty damn good. Arizona actually finished 2nd in both of those two categories. He’s certainly not losing anything with his new players behind him.
Most of that is because a) They have a good infield defense, and b) they have micro-sized ballpark. The single biggest weakness of defensive metrics (other than not knowing squat about catching) is its strong preference for outfielders in small ballparks.
While that may be true, it doesn’t exactly change the fact that playing for CIN can partially explain why his BABIP is below league average. Arizona plays more hitter-friendly than GAB in every way except for homers, so while Arroyo may give up fewer homers in Arizona, his BABIP may very well go up as well.
I’m don’t have the numbers to back me up right now, but I am almost certain that the Diamondbacks have a better outfield defense that the Reds. Trumbo will probably not be a very good defender, but I am pretty sure that Pollack and Parra were both top 10 in UZR last year.
I believe Arroyo has the beat your FIP skill and his age doesn’t necessarily bother me. The big red flag for me is his velocity.
He’s basically at the bare minimum that non knuckle ball righties can throw and still survive. Whenever he loses a tick or two, he goes into that death spiral of giving up more homers which leads to less thrown strikes which leads to more walks which leads to more 3run dingers which leads to the end.
To be fair though, his velocity has plateaued for 4 years.
Yeah, I’d say he already lost his tick or two, and should remain at this level for another couple of years.
You hit the nail on the head. He’s a 37 year old pitcher who gives up bombs and can’t strike people out and he’s now playing at Chase Field. I’m not sure how you can get excited for this if you’re a D-Backs fan.
You can “hope” he keeps over-performing but he’s working against the age curve and has no velocity. Even if this is a depth signing, it’s going to be an overpay.
Yes, we would expect x number of pitchers to outperform their peripherals based on randomness alone. But we can see that there are many more pitchers who outperform their peropherals (at least BABIP) than randomness would expect.
http://www.insidethebook.com/ee/index.php/site/comments/career_dips_numbers/
So we know BABIP skill exists for many. So what does that mean for Arroyo in particular? Regress his BABIP-we know what point r=.5. That should be our best guess as to his BABIP skill.
As far as LOB% – there is a lot less information about the point in which r=.5. I have seen it around that it takes maybe 4000-5000 PA for r=.5. For sequencing events. So we should be reaching the point that we can get a solid estimate of his LOB% skill, if he has any.
We know Glavine is the famous prototype for giving up fewer HR with men on base than with nobody on. I would like to see some sort of comprehensive study going back as far as data allows to see how common this approach was/is.
Oh. Forgot something connected to what was mentioned earlier. On the point on ERA vs. FIP. Tom Tango also suggested that it takes about 6 200 IP seasons for RA9 to become a better indicator of pitching talent than FIP.
That’s for a sample of pitchers as a whole. If you agree to the idea that some pitchers can control their BABIP to some point, its a better indicator much faster.
Also, why are we still not using SIERRA here? It’s leagues better than FIP.
The thing that bothers me about SIERRA, etc. is the batted ball profile assumptions. SIERRA would have Greg Maddux, for example as a below average BABIP pitcher since he was an extreme GBer. It makes no corrections for the fact that extreme GBers have better BABIP on GB than other pitchers…and other things like that.
I should clarify. I think, like FIP, SIERRA works for a bast majority of pitchers. I just am not sure outside of picking-up on FB vs. GB tendencies that it will do much to help us with cases like Arroyo much more than other tools. It still likely isn’t discovering how Arroyo is “doing it”, if he is “doing it.”
I can’t believe you tards think Bronson Arroyo is a good pitcher or that he’s going to have a good season. Guy is hot garbage that has been getting by in a terrible division. Put him in the AL and you’ll have to put him in the pen by July. This is a hilariously bad deal, but don’t let the obvious get in the way of your argumentative nature.