Top 10 Overrated Starters in 2009
We're at the point in the baseball season where all the two-week flash-in-the-pan performances have melted away and nobody is on pace for 100 home runs anymore. But that doesn't mean all the numbers out there are for real, especially on the pitching side of the ball.
Some starting pitchers with pretty early-season ERAs are for real, and some aren't. To separate the two, we need to look at peripherals, most importantly strikeouts, walks, and homerun prevention. One nice estimate of a pitcher's demonstrated talent level is FIP, which takes those three skills and produces an ERA-like number on the same scale. I like to think of FIP as "the ERA a pitcher deserved given the skills he demonstrated."
To produce the list of the ten most overrated starters (based on 2009 ERA) below, I went to Fangraphs, selected all the pitchers who had thrown at least 25 innings as of Sunday's games, and found the ten whose ERAs were the lowest relative to their FIPs (E-F). Also listed is tERA, Stat Corner's advanced DIPS-style metric, as a backup to FIP.
- Jair Jurrjens (Braves) -- 40.1 IP, 2.01 ERA, 4.34 FIP, -2.34 E-F, 3.97 tERA
It's not that Jurrjens has been bad, it's just that he's not a stud. Striking out 4.5 hitters per game and walking 3.5 per game is not a recipe for sustained success. He was better than this last year, and should be an average to slightly above average starter going forward. Just don't think he's found the secret to future ace-dom. - Matt Cain (Giants) -- 38 IP, 2.61 ERA, 4.92 FIP, -2.31 E-F, 4.47 tERA
Dan Szymborski and I talked about Matt Cain a bit in our chat yesterday. Yes, he's been historically unlucky with run support and his W-L record. But the Giants also play in a stadium that's friendly to fly ball pitchers, and Cain's definitely a fly ball pitcher, meaning he's also lucky in some regards. Anyway, giving up over one home run per game and sporting a 1.5 K/BB doesn't support a sub-3.00 ERA. His career hom run rate is a bit better, but expecting an ERA much lower than 4.00 the rest of the way is a stretch. - Trevor Cahill (Athletics) -- 33 IP, 3.82 ERA, 5.85 FIP, -2.03 E-F, 7.33 tERA
No, that tERA isn't a typo. Cahill has a ton of talent and hitters aren't doing anything with his pitches when he puts them in the strike zone. Unfortunately, he hasn't been able to find the strike zone this year, walking just under five batters per game. Without any changes, he'll be out of the rotation soon. However, if (and that's a big if, at least in the short term) he can put more balls over the plate, he would see his <1 K/BB rate rise, helping to offset the increase in BABIP he's likely to see. - Doug Davis (Diamondbacks) -- 44.1 IP, 3.25 ERA, 5.23 FIP, -1.98 E-F, 5.14 tERA
Nobody can root against Davis, but giving up 8 home runs usually results in more than 16 runs allowed. Going forward he'll probably give up fewer long balls and revert to the 4.25 ERA pitcher he's been for the past couple years. - John Lannan (Nationals) -- 39.1 IP, 3.89 ERA, 5.77 FIP, -1.88 E-F, 5.99 tERA
It would be nice to say the Nationals have a bright spot on the pitching side of the ball, but 20 Ks against 15 BBs and 7 HRs isn't even a glimmer, let along anything bright. - Kevin Millwood (Rangers) -- 52.1 IP, 2.92 ERA, 4.63 FIP, -1.71 E-F, 5.16 tERA
I root for Millwood, and walking under two hitters per game is impressive, but a Carlos Silva-like strikeout rate leaves too many balls put in play, especially ones that go over the fence. As his tERA is right in line with what both his tERA and ERA were the past two seasons, an ERA around 5.00 is the most likely scenario the rest of the way. - Chris Volstad (Marlins) -- 42.1 IP, 2.98 ERA, 4.67 FIP, -1.69 E-F, 4.90 tERA
Like the Marlins win-loss record, Volstad's early season numbers are a mirage. A .231 BABIP on a team that doesn't field well in general isn't a sign of ERA sustainability. His K/BB rate is good and if his 17% HR/FB rate can come down a bit, he'll be an above-average starter, just not a Cy Young contender. - Jered Weaver (Angels) -- 40.2 IP, 2.66 ERA, 4.31 FIP, -1.65 E-F, 3.42 tERA
Here's a case where tERA disagrees with FIP by almost a full run, most likely because FIP regresses Weaver's .230 BABIP all the way, while tERA likes the fact that 18% of his batted balls are infield flies. His 7/2 K/BB ratio is excellent, although because he's giving up more fly balls overall, his HR/9 may rise a bit, meaning one should expect good, but not great, stats the rest of the way. - Mark Hendrickson (Orioles) -- 26.1 IP, 5.13 ERA, 6.75 FIP, -1.62 E-F, 7.09 tERA
When your ERA is over 5.00 and you've been lucky, that's not a good sign. You don't need me to tell you that Hendrickson won't be a serviceable pitcher any time soon. The only good news for Baltimore fans is that the faster Hendrickson's ERA flies north, the sooner they'll start to see the glut of Oriole pitching prospects called up to the majors. - Joe Saunders (Angels) -- 47.1 IP, 2.66 ERA, 4.23 FIP, -1.57 E-F, 3.05 tERA
Saunders' ERA was well below his FIP for the 2008 season as a whole, and he's doing it again this year. Like another Angel above, his tERA looks much better than his FIP, thanks to a 23% IF/FB. Your choices are to think he can sustain producing an enormous percentage of batted balls that are infield flies, or being worried that a 4 K/9 will eventually catch up with him. I'll take option two. I do wonder, however, if there's a chance the Angels have an organizational philosophy towards inducing infield flies...
Here are some quick hits without much commentary. I looked at the next fifteen or so players based on ERA-FIP and picked out those whose peripherals were the most scary: Mark Buehrle (decent K/BB indicates he's more of a high 3.00's ERA guy), Lance Cormier (the only reliever with 25+ innings is striking out less then three batters per nine innings and has given up home runs on only 3.7% of fly balls), Tim Wakefield (yes, knuckleballs are hard to hit, but his .228 BABIP and .23 HR/9 rates are bound to go up and will no longer hide a 1.4 K/BB ratio), Scott Richmond (his peripherals are average across the board, enough said), Aaron Laffey (more walks than strikeouts and only 4% of fly balls have gone for home runs), Roy Oswalt (what happened to all the Ks?), and Zach Greinke (his FIP is a lofty 1.46).
53 comments
|
0 recs |
Do you like this story?
Comments
wakefield
its just so awesome that people still think wake is gonna deliver something other that EXACTLY league average pitching over the course of a season.
tra+ since 2003 (100 being league average): lowest 96, highest 103
era+ average since 2003 (again 100 is league average): 106
so wake gives about 180 ip of league average pitching for $4mil per year (plus some performance bonuses) thats VERY cheap
Wakefield's contract deserves more discussion in the general realm of baseball than it gets.
It’s such an incredible anomaly and I think it’s pretty amazing.
Future Redbirds - tracking Cardinal prospects for Cardinal Nation
seriously
shouldn’t Wakefield’s agent have his license yanked?
Do Boston fans actually complain about that contract? Yikes. Move over Yankees fans…
I'm not a sabermetrician, but I do play one at Driveline Mechanics.
by Matt Klaassen on May 13, 2009 3:00 PM EDT up reply actions
Not about the contract
they complain about his pitching in general and the fact that he requires his own catcher. Especially this past offseason when our catching positions have been in limbo. It’s pretty awful that some of the Nation can’t realize that per dollar he is helping the team more then anyone.
Agreed.
Although NG always complains about Wake, most of them thought he wasn’t worth the $4 mil – insane.
How do you convert tRA to tERA?
Do you just subtract 0.4?
"Good pitching will always stop good hitting and vice-versa."
-Casey Stengel
I'm subtracting .35 here.
Sometimes I divide by .92, which is probably a bit more accurate.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
by Sky Kalkman on May 13, 2009 12:16 PM EDT up reply actions
Where do you get the number .4 from? Just curious.
by .Taylor on May 13, 2009 5:19 PM EDT up reply actions
It's the average number of unearned runs per nine innings
Or the difference between the league average RA and ERA.
The reason .92 is better is because good pitchers will give up fewer unearned runs per game than bad pitchers. So using .35 for good pitchers is too much and their tERAs will look a bit too low adjusting from tRA, and vice versa for bad pitchers. For work like this, it doesn’t really matter.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
That's interesting. tERA isn't on StatCorner, sadly.
Do you use the previous year’s average number of unearned runs?
by .Taylor on May 13, 2009 5:49 PM EDT up reply actions
.35 and .92 are just numbers that are stuck in my head. I could (should?) be more precise.
Eh, the margin of error on that isn’t as large as FIP anyways.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
No, I'm not questioning you, haha! I just was curious if the averages used to formulate tERA and FIP, I suppose as well, vary from year to year.
by .Taylor on May 13, 2009 5:58 PM EDT up reply actions
FIP varies, sort of.
The formula is the same, but includes adding a constant to make league average FIP equal to league average ERA. It’s a fudge, but a remarkably good one.
I’m not totally sure about averages for tRA. I’m pretty sure the statcorner guys make league average tRA equal to league average RA each season, but I don’t know if they adjust the weights on each batted ball event each season or handle it like FIP with an overall adjustment and simply use event weights averages over many seasons worth of data.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
For a list of FIP constants for each league 1871-2008 I did a few months back
click here
The list includes lgERA and lgRA for each season, and the FIP constant for scaling FIP to either ERA or RA
The fomula for FIP this is based on is
((HRx13 + (BB-iBB+HBP)x3 + SOx2)/IP) + Constant
I know that’s easy to find, I just wanted to note what my assumption was since different people handle the iBB and HBP in various ways.
I'm not a sabermetrician, but I do play one at Driveline Mechanics.
by Matt Klaassen on May 13, 2009 6:51 PM EDT up reply actions
I don't consider HBP and IBB
In the end, they pretty much even out.
St. Louis Cardinals... defying win expectancy since 2008
by vivaelpujols on May 14, 2009 12:08 AM EDT up reply actions
Matt Cains K/BB rate is always bad his first month and half of the season. Just look it up from the last 3 seasons. He always improves as the season goes along.
You look it up ;)

I guess. He certainly finished strong in ’06 and ’07, but not ’08. And early ’06 was right at his seasonal average. The shape of the ’08 curve is a flip of the prior two.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
by Sky Kalkman on May 13, 2009 12:33 PM EDT up reply actions
What Millwood is doing to make Rangers fans happy is going deep in to games.
He’s good enough that those extra innings mean a lot, especially in helping them avoid middle relief and keep the bullpen fresh.
But his ERA in part is indication of how much the Rangers’ defense is helping out their pitching. It’s the major reason for the improved run prevention thus far.
Wouldn't some of the reason Millwood's going deeper into games be that he's getting lucky on BIP outs?
And for the same reasons his ERA will rise, he’ll get fewer outs per pitch and need to be yanked earlier?
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
Yeah, he's thrown at least 11 pitches in 6/7 starts.
http://www.baseball-reference.com/players/gl.cgi?n1=millwke01&t=p&year=2009
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
at least 11 pitches in his starts?
Am I missing something or should that be 111?
If you were thinking, you wouldn't have thought that.
As an example, look at statcorner's Outs minus ExpectedOuts (O-xO)
Millwood’s at 15 this year, meaning he’s gotten 15 more outs than you’d expect based on the amount of balls in play and their batted ball type. 15 outs is five innings. Over seven starts, that’s just over half an inning of “luck” per start or, say, eight pitches. Not huge, but it’s there.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
FIP
I really love FIP, it has many useful purposes including comparing pitchers on different teams to see how much defense makes a difference. On the flip side I don’t think FIP is the perfect statistic for determining how “lucky” someone is. I think FIP is skewed and doesn’t compensate for “groundball” pitchers. Wakefield is not a ground ball pitcher but he certain gets alot of balls put into play that FIP does not compensate for.
Looking at Wake’s numbers in 17 seasons he has a career 4.30 ERA and 4.72 FIP. That is a difference of .42, around 10% of his ERA with a large sample size. Has Wake just been lucky for 17 seasons in a row?
Lets look at some great groundball pitchers of our time. Brandon Webb over 7 seasons has a career 3.27 ERA and 3.50 FIP. Tim Hudson over 10 seasons has a career 3.48 ERA and 3.79 FIP. Those are differences of .23 and .31 respectively with fairly large sample sizes.
Are we to say that these pitchers have been lucky their entire careers or that FIP isn’t the end all say all of a pitchers contribution? I would say the latter could be the case in some situations, we just don’t know.
Does any of this have to do with the types of pitcher's we're dealing with?
Like Webb, Hudson, and Cahill earlier. Does FIP take into account that they were groundball pitchers and induce more double plays than the average pitcher? As an A’s fan, I’ve seen most of Cahill’s luck coming from DPs to get out of jams. I know he can’t always come up with that double play, but I don’t really know how this would affect his FIP—I’m not a genius when it comes to the the more advanced pitching stats.
FIP accounts for GB/FB and GIDP indirectly, not directly.
FIP isn’t perfect, both in its construction and in the validity of its assumptions. Pitchers DO differ in ability to control many things we assume they don’t differ in. It’s just that nobody’s come up with ways to detect those, as of yet, except for using larger and larger sample sizes.
In general, differences of 1/3 of a run don’t mean much to me. 2/3 of a run and I start expecting change. More than that and I’m sure about it.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
Good points, definitely.
There’s research that knuckleballers have significantly lower BABIPs. Not as low as Wakefield this year, but lower. Let’s say that .42 FIP-ERA is a constant for Wake; his 4.10 FIP this year would drop to about 3.70. (His ERA is 2.93.)
Regarding GB/FB pitchers, FB pitchers have lower BABIPS, but have a higher SLGBIP (fly balls are caught more often but when not caught turn into extra base hits more often.) The saber-cliche is that those effects cancel out, but I’m sure it’s not exact.
With Webb and Hudson specifically, think about their parks and defenses. Hudson’s A’s years were in a pitcher-friendly park and with good defenses, I’m guessing. Are his ERA-FIP numbers better in OAK than with ATL, perhaps. Webb’s been in a strong hitters’ park and his defensive haven’t been anything special. Of course, even over ten years you’ll see differences of .30 runs of ERA. That’s not to say there’s nothing there, just that it doesn’t prove anything. A large-scale study would be awesome to see.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
I have some issues with FIP and Tango helped me with a verson between FIP and tRA
(I don’t like tRA because of inconsistencies in calling line drives)
Basically it uses only FB rate (bad – it leads to HRs) and GB rate (average defense get 70% of them for outs). I like to use it on the off season because BB, SO, FB and GB are projectible from one season to the next.
by Jeff Zimmerman on May 13, 2009 5:05 PM EDT up reply actions
If you use both FB% and GB%, aren't you inherently using LD%?
But yes, I like constructs like that better. Doing this again, I might use xFIP instead of FIP. It still misses the HR piece because it regresses HR/FB 100% towards 11% instead of 0%, but it probably misses by less.
BPro often uses something they call quick ERA which is based on K, BB, and GB%.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
Those 3 is what I used originally until Tango chined in
(SO – BB) + (.7)*GB
Gives the number of at bats that parks, wind, luck isn’t involved
by Jeff Zimmerman on May 13, 2009 5:33 PM EDT up reply actions
So are you willing to share the formula you use or are you going to sell it to the highest bidder?
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
Its been public knowledge for a while
Wrote that article on my own before I got sucked into the BtB cult.
by Jeff Zimmerman on May 13, 2009 6:13 PM EDT up reply actions
Ah, so that's the formula above, got it. I was expecting something on the ERA scale.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
Why does everyone have to reference pitching numbers back to ERA
Why can’t they stand on their own?
by Jeff Zimmerman on May 13, 2009 6:27 PM EDT up reply actions
perhaps easier conversion to runs above/below average?
I'm not a sabermetrician, but I do play one at Driveline Mechanics.
by Matt Klaassen on May 13, 2009 6:53 PM EDT up reply actions
Because then you have to adjust to a different scale.
If you always present players in a list, then you have relative reference, yes. Or you can tell me the rank of a player’s Quatlu to add context. But if you tell me Joe Mesa’s 2008 Quatlu is 17.8, what the heck does that mean?
If you instead put new metrics on scales people are familiar with, we know a good Quatlu when we see it and the metric becomes more popular.
Additionally, with something like FIP, you can use it to project future performance quite easily. I’m sure you could project future PPQ with past PPQs and then convert that to ERA for an ERA projection, but why not just do that to begin with?
The big exceptions to converting a new metric to an old scale are when it has meaning on its own or is on a familiar scale (like 0 to 1). For an example of the second one, strike zone percentage is strikes_in_zone/pitches. I might not know what the league average is, but I understand a percentage.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
Mark Hendrickson
is no longer overrated, he just stinks after last night’s horrorshow (2.0 IP, 6H, 5ER, 1 BB, 0 K)
ERA 6.35
FIP 6.61
Bring on the cavalry!
"Honestly, I get tired driving 26 miles. I couldn’t imagine running it." -- Huffy
Oh, but tERA says he's still got some room to fall!
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
I've watched him pitch
and I’m not surprised. He is a hit machine. BAA = .345
If he wasn’t tall and didn’t throw with his left hand, he’d be out of a job last year.
"Honestly, I get tired driving 26 miles. I couldn’t imagine running it." -- Huffy
I have a question about FIP
It appears to me that it is basically linear weights (although I could be wrong), however it looks like it treats each event (HR,BB,SO) independently, which doesn’t make sense at all.
For example, Cole Hamels gave up a lot of home runs last year. However, because he limits base runners so well, due to his high K rate and low BB rate, his home runs, on average, wouldn’t be as detrimental as they would be to a pitcher with average K and BB rates. However, it appears that the FIP formula treats Hamels’ home runs the same as it does Daniel Cabrera’s, when Hamels’ are clearly “better”.
I would assume that Tango would have though of this (hell, even I did), so it might be that I am just not understanding the FIP formula well. Could Sky or someone else please clarify things for me?
St. Louis Cardinals... defying win expectancy since 2008
FIP has issues, that is why you will read some people using FB rates (my comments above - which Tango states is better) or xFIPS
Philly’s Stadium is a launching pad and a higher than normal amount of FBs will be HRs. If you goto the Hardball Times, they used xFIPS, which uses something like every 1 out 9 FB will be a HR (don’t quote me on that), so pitcher in San Diego (pitcher’s park) that gives up the same number of FB is rated the same as one in Philly.
If you take it one step further, to tRA, linear weights for pitchers, there is a huge problem with scorer errors, so some people have problems going to it.
by Jeff Zimmerman on May 14, 2009 1:00 AM EDT up reply actions
Yeah I know about xFIP and tRA
I think that they are slightly better than FIP, but are much harder to calculate, so I don’t use them as much. However, those stats still suffer from the same problem that FIP does IMO, which is that they apparently treat each event independently, when they clearly shouldn’t be.
St. Louis Cardinals... defying win expectancy since 2008
by vivaelpujols on May 14, 2009 1:07 AM EDT up reply actions
No, the BB, K, and HR terms don't directly interact.
But the whole reason for including Ks in the formula is to indicate that fewer batters are reaching base by hits. In the grand scheme of things, a K and a GB out are not significantly different. But when a pitcher Ks a lot of batters, that means fewer balls in play and fewer BIP hits. So Hamels’ 10 HRs in the FIP formula count against him as much as Carlos Silva’s 10 HRs, but then the K piece indirectly reduces the the value of those HRs.
You can watch FIP derived here from linear weights, although you might say HRs aren’t interacting with anything there, either.
You familiar with BaseRuns? It’s a run estimation formula that looks like (RunnersOnBase-HR)*(ScoreRate) + HR. Makes sense, right — you either get on base and get knocked in sometimes or you’re the batter who hits a homerun You can expand that to (and I’m leaving out some details):
(H – HR + BB) * (good things/(good things + bad things)) + HR
or
(H – HR + BB) * (H + BB)/(H + BB + outs) + HR
You can go all DIPS on it by replacing H and outs with K and (AB – H – K) * .300 and adding appropriate weights to the various events. In fact, Colin did.
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
If people are interested in a SQL for converting COlin's BsR-FIP to WAR, I wrote one
Should have a short post up over at Driveline a bit later that uses is just a bit. Very interesting results.
I'm not a sabermetrician, but I do play one at Driveline Mechanics.
by Matt Klaassen on May 14, 2009 11:57 AM EDT up reply actions
okay, here it is
I'm not a sabermetrician, but I do play one at Driveline Mechanics.
by Matt Klaassen on May 14, 2009 1:02 PM EDT up reply actions
I'm going to try to resist fanshotting it
I'm not a sabermetrician, but I do play one at Driveline Mechanics.
by Matt Klaassen on May 14, 2009 1:03 PM EDT up reply actions
Why?
If it’s good, do it. If not, uh, I’m sure it’s good…
Beyond the Boxscore // Calling BJ Upton lazy is lazy.
thanks, I will
I just don’t want to flood fanshots with self-promotion
I'm not a sabermetrician, but I do play one at Driveline Mechanics.
by Matt Klaassen on May 14, 2009 1:44 PM EDT up reply actions
Cole Hamels Belongs
Hey Sky, this top ten list is fantastic. I vote for Cole Hamels, this is supposed to be an ace who has an ERA of 6.17, pathetic. You can post this to our site http://www.toptentopten.com/ and link back to your site. We are trying to create a directory for top ten lists where people can find your site. The coolest feature is you can let other people vote on the rankings of your list.

by 






























