FIP is a Garbage Statistic
Everyone seems to argue that ERA is biased in favor of pitchers with good defenses, and while that is true, it is still the best way to measure the quality of a pitcher; at least a starting pitcher. I will argue that for a relief pitcher, WHIP is more important since you often want a guy that can get out of a situational jam (IE: Bases loaded, 1 out), but I digress.
As most of you know, FIP stands for "Fielding Independent Pitching," and is a variation of sabermetrics. The formula is as follows: FIP = (13 HR + 3 BB - 2 K) / IP) + 3.1.
This formula FAILS because it implies that all successful contact pitchers are lucky. It can take a great pitch to jam a good hitter into hitting a pop up infield deep, and it can take a great pitch to have a good hitter tap a routine ground ball. None of these are rewarded with this formula.
You can make bad pitches which go for screaming line drives, doubles/triples off the wall, ground balls up the middle, etc. All of which even a good defense wouldn't save you from. None of these are penalized with this formula.
It doesn't take a baseball genius to see that this formula rewards pitchers that pitch for strikeouts and avoid homeruns. The pitcher that pitches for contact is screwed. The pitcher that gives up a lot of base hits, gets a lot of strikeouts, and avoids homeruns benefits. I'm not saying ERA is perfect. I'm just saying it's better than this garbage formula which has several biases of its own.
It only took me one attempt to think of a pitcher that could disprove this formula. James Shields of the Tampa Bay Rays. I'm sure everyone here agrees; the Rays have one of, if not the best defense in the league. Despite giving up a lot of homeruns, James Shields gives up very few walks and has a lot of strikeouts. His FIP is 4.099 while his ERA is 4.79. I guess the Rays' terrible defense has really let him down.
And don't get me started on why contact pitchers are better than strikeout pitchers. That's a whole different argument.
42 comments
|
0 recs |
Do you like this story?
Comments
This was my reaction.
In baseball the object is to go home! And to be safe! "I hope I'll be safe at home!"
-George Carlin (RIP)
As someone who has watched Halladay over the years, and seen him pick apart a batter to point where they, after 4 or 5 pitches, finally make a bad swing and produce an weak grounder to the inield, I can be aboard with this. Halladay has only recently become a K pitcher, but his first few year, especially when he won the Cy Young award, he was a pure ground ball pitcher. Luck had little to with anything.
There She Gooooooes!
Ummm . . .
There are too many things wrong with this to count . . . I will just note that when Doc won the Cy, his K/BB was a league-leading 6.38 — he wasn’t a high-K pitcher (and still isn’t, though he’s been over 7 per 9 the last few years, which is a little higher than the 6.9 he managed that year), but he barely walked anybody. And yes, he was at the time a strong groundballer, which is a good thing — less because it produces outs (they aren’t all “weak grounders to the inield” [sic] by any means, and even some of those find holes) than because it avoids extra-base hits.
Luck is always a factor, and you clearly don’t understand the actual arguments.
by The Ancient Mariner on Aug 1, 2010 11:34 AM EDT up reply actions
Wow.
Have you read ANY of the relevant studies? Voros’ DIPS study, maybe? No. But you know that James Shield’s “disproves” the fundamental concept.
Again – wow.
by cwyers on Aug 1, 2010 9:57 AM EDT reply actions 3 recs
Okay, Rays125, take a step back. Do you honestly think that thousands and thousands of educated, intelligent people haven’t gone through the same thought process you have? Do you honestly think that no one has dared to ask the questions you’re asking? Because everyone has. And we’ve gotten answers. Instead of sitting back and looking at random handpicked examples, how about you take a step back and why this formula came to be.
Firstly, pitchers are very inconsistent on their batting average on balls in play (BABIP). While strikeouts have a year-to-year correlation of 0.7, BABIP is closer to 0.1. There are some differences between pitchers in terms of BABIP skill, as many have studied and discovered, but just not as much as the difference between pitchers in strikeout skill.
The first thing to know about FIP is that if you take all of the FIPs and ERAs for pitchers in 2009, the 2009 FIPs will be closer to the pitcher’s 2010 ERAs than the 2009 ERAs will. The reason is that you look at the most persistent statistics when you use FIP, and you get rid of the statistics that aren’t much under a pitchers control. It happens every year, too. The FIPs on average are always closer to the next year ERAs than the previous year ERAs. Every single year.
How about you take a step back and read the relevant articles? Thousands of us have and have understood that contact pitchers are only good if they are the type that avoid walks, and the type that avoid home runs. Contact pitchers are very singles-prone, so preceding those hits with base on balls is a big problem and following them with home runs is a problem.
Once you read all of the articles that people have written looking at this stuff, you would realize that it can help you understand baseball a lot better.
by Matt Swartz on Aug 1, 2010 11:29 AM EDT reply actions 4 recs
FIP (and to a greater extent xFIP) does take into account being a contact pitcher.
It just says that if you want to be a contact pitcher, you have to limit your walks and get groundballs (and therefore limiting home runs) since you aren’t going to be striking many people out. I think that is a pretty common sense formula that most contact pitchers that succeed have followed.
by ThePanda on Aug 1, 2010 12:00 PM EDT reply actions 1 recs
This guy isn't worth arguing with.
Just a casual fan that got pissed because he doesn’t understand sabermetrics.
You give Rays fans a bad name with your lunacy.
Don’t lump all of us in with this Martian.
Not my cat. My cat isn't fat like that. I use to have a fat maine coon though.
2010 Trade Deadline Primer --> http://dockoftherays.com/2010/07/03/2010-trade-deadline-primer/
Be easy on newcomers
I believe everything has been said. I think what he means is that while FIP has its uses, and a good predictive value, it’s hard to come up with statistics that can measure the past more accurately by eliminating factors.
And not every place on the internet is as postmodern as this one when it comes to discarding traditional arguing techniques. And having any fan interested in any form of serious analysis is good.
If you struck nobody out, walked nobody, and gave up no HRs, FIP would say your ERA should be abotu 3.20
That’s really really good for the epitome of a contact pitcher.
You certainly have to appreciate that he decided to try to argue against compontent-based ERA-scaled metrics on this site
But as others have said above, it just seems like the writer of this post doesn’t entirely understand WHY these other metrics (FIP, xFIP, tERA, SIERA, etc.) are often more reflective of a player’s individual performance than ERA.
It’s too bad that he’s not really participating in the discussion that he managed to create, though, because a lot of his problems with FIP could probably be assuaged by learning from some of the very smart people who comment here.
I like baseball.
I write for Beyond the Box Score and The Hardball Times Fantasy
Guys.
http://www.sbnation.com/users/Rays125
Check out the new FREE online baseball game! I know you'll love it.
Two Out Rally, BASEBALL MMORPG | Facebook | @2OutRally
Well, since mu question probably isn't worth an entire fanpost
and it’s about FIP, I’ll just throw it in here. But here it is: do pitchers on teams with better defenses trend towards having better or worse FIP’s or is there no correlation? Same for any other stat that attempts to completely separate pitcher performance from fielding. The reason obviously being that we would prefer no correlation (I assume).
All is vanity.
I don't think this even needs to be checked for, presumably.
Just logically speaking there’s no connection between BB, K, and HR and fielding aside from the very rare home-run saving plays.
Just at a glance through the last few years (not very scientific, I know), the only time there’s been any correlation is at the bottom this year with a few teams, and that could be a combination of random chance and the fact that, quite frankly, there are some really badly put together teams out there.
USG
by Ben Buchanan on Aug 4, 2010 12:36 AM EDT up reply actions
That would assume pitchers pitch exactly the same no matter what their defense is like. This is not true, as pitchers with better defenses admittedly pitch more to contact. I would guess there is a small correlation.
I agree
But it’s a small effect:
http://www.3-dbaseball.net/2009/11/pitching-to-contact-and-fip.html
Yea, I think you're discounting how much pitchers might change their approaches based on who's behind them.
A similar example would be park factors for Ks and BBs, even though it intuitively makes sense that the park wouldn’t effect these stats, they do, and sometimes to a pretty large degree.
it intuitively makes sense that the park wouldn’t effect these stats
Disagreed. Every park has a different hitting backdrop, different shadows, different air conditions (pitches move differently based on temperature, humidity, and other weather conditions), etc.
by Missing Barry on Aug 5, 2010 10:42 AM EDT up reply actions
FIP is scaled based on K/9 which equals (K / IP) * 9
IP is influenced by BABIP, which is influenced by defense. A pitcher that faces 300 batters with a .300 BABIP and 50Ks and 0 BBs would have a line that looks like this:
50 K, 0 BB, 75 IP, FIP = 1.87
Adjust that pitcher’s BABIP to .325, with 300 batters faced and 50 Ks (same K%) and you get:
50 K, 0 BB, 73 IP, FIP = 1.83
The “worse” pitcher by luck, gets a higher score because he got the same K rate in fewer innings because of more hits behind him.
Alternatively, make the BABIP .275 and you get a worse pitcher:
50 K, 0 BB, 77 IP, 1.90 FIP
Greg Maddux, 1995. BABIP = .253, TBF = 785, K = 181, BB + IBB = 26, HR = 8, IP = 209 2/3
Standard FIP = ((181*-2) + (26 * 3) + (8 * 13) ) / 209.667 + 3.2 = 2.34
Expected BABIP = .300
Expected Outs = 181 + (785-181-8-26)*.7 = 580 = 193.3 IP
Modified FIP = 2.27
(2.34 – 2.27)/9 * 209 2/3 = 2 run difference in valuation (.2 wins)
So, differences in BABIP using the standard FIP calculation between identically performing pitchers will result in different FIPs based on the BABIP (and other defensive factors) behind them. But, those differences are quite minor — Maddux with a ridiculous BABIP over 210 innings still resulted in only a difference in evaluation of 2 runs.
Matt Swartz:
Thanks for your intelligent response. While you make a strong argument, I still disagree.
I agree that FIP is successful in measuring ONLY that which a pitcher can control. My argument remains that those things aren’t necessarily important. The only factor that FIP uses in its measurement that I consider significant is walks. You can be a bad pitcher and not give up homeruns, and you can be a great pitcher while not getting many strikeouts. At the risk of repeating my original post, I won’t elaborate any further. Just keep in mind that a light dribbler to the third baseman is going to be an out even in Single A.
Seemingly, the strongest point you make is that FIP is a better predictor of ERA for subsequent years than ERA from prior years. That said, it doesn’t really support either of our claims. In order for your claim to be true, ERA would have to be an invalid measure one year and a valid measure the next. You guys all say ERA is an invalid measurement, so to claim FIP is a good predictor of ERA is to say that FIP is a good predictor of an invalid measure. That doesn’t make much sense.
Past ERA is an invalid predictor of future ERA, not an invalid measure of performance. A pitcher’s value is determined by the number or runs he gives up, so that is why ERA is predicted.
HS team nickname: Redmen, College team nickname: Warriors, Amateur team nickname: Chiefs, Favorite MLB team: Braves. Holy political incorrectness...
by LeeTro on Aug 4, 2010 5:22 AM EDT up reply actions 1 recs
Yeah, this is the point. The goal of pitching is basically to prevent runs from scoring. So, ERA (or really RA, including unearned runs, but that’s a relatively minor issue) would be the best measure of output. But just like Runs Scored is a pretty weak measure of how good you were on offense (after all, it’s kind of important who was behind you in the batting order and who was behind you in the field as a pitcher), Runs Against is a pretty weak measure of how good you were on the mound out of context.
So predicting who has the talent level such that they are likely to give up the fewest runs (adjusting for context) is really the goal. But the point is that you can predict who is likely to give up the fewest runs by looking at the statistics that repeat themselves. K-rate, BB-rate, and GB/FB are the rates that repeat themselves year in and year out. How many singles the pitcher gave up last year seems to correlate poorly with how many singles the pitcher is going to give up this year. So if you look at what his ERA would be with a pretty average amount of singles will tell you about how well he is likely to do next year.
Again, you need to read a lot more before continuing on with this.
These are all old arguments.
Some links:
http://www.baseballprospectus.com/article.php?articleid=878
http://www.baseballthinkfactory.org/files/primate_studies/discussion/lichtman_2004-02-29_0/
http://ussmariner.com/2006/08/29/evaluating-pitcher-talent/
http://web.archive.org/web/20080112135748/mvn.com/mlb-stats/2008/01/06/on-the-reliability-of-pitching-stats/
and maybe especially Kincaid’s stuff:
http://www.3-dbaseball.net/2009/10/evaluating-pitchers-with-fip-part-i.html
http://www.3-dbaseball.net/2009/10/evaluating-pitchers-with-fip-part-ii.html
http://www.3-dbaseball.net/2009/11/pitching-to-contact-and-fip.html
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
I've had some of the same issues with FIP
But it’s not that it’s a junk stat, but rather that it will overrate some pitchers and underrate others. While on average BABIP doesn’t have much of a correlation, it is a repeatable skill for certain individuals. Thus, you end up with some guys like Andy Sonnanstine (who has a higher than average BABIP year in and year out, this SSS year notwithstanding, for what I think are pretty obvious reasons) who get overrated by FIP, and some others like Armando Galarraga (off the top of my head), who seems to be able to pitch to weak contact, who get underrated. A guy who has some movement and deception on his pitches, but not enough to generate swings and misses at a high rate, probably gets hurt a bit, since his pitches are likely to be hit less solidly.
But if one can reason that a pitcher has a certain amount of control over whether his pitches miss bats, then one could also reason that a pitcher should also have a certain amount of control over whether his pitches partially miss bats, inducing a lot of foul balls and weak contact.
All that said, FIP certainly has its uses (and on the balance is probably a better stat than ERA), and definitely has much higher predictive qualities than ERA.
Bad Left Hook - The SB Nation boxing blog
"Baseball is played on the field, not on a calculator."
I also find it funny that a guy named "Rays" posted this
If this had been posted at DRaysBay, it would be on the favorites list due to massive ironic reccage.
Bad Left Hook - The SB Nation boxing blog
"Baseball is played on the field, not on a calculator."
I think that what RaysFan125 is trying to say
is that FIP does not take into account the kind of balls that are hit back. It only takes into account the things that the final result is in the pitcher’s control, which are BB’s, K’s, and HR rate. Now, admittedly, he phrased it in a bit of a confrontational way and I think thats why everyone’s up in arms, but he has a valid point. Pitchers do have control over at the very least, GB rate, which pitchers are consistent in year-in and year-out with (Brandon Webb has never suddenly given up all FB’s in a season), and from the limited amount of knowledge I have, there is an argument whether pitchers can control pop-ups as well (I recall Jon Rauch being noted as particularly adept an inducing them). So, I don’t think FIP is completely a “garbage stat” as it is pretty useful in evaluating pitchers minus their defense, I dont think it tells us the whole story, like Raysfan125 is saying, if I’m getting him right. I think in order to truly get at a pitcher’s worth, you have to look at the various batted ball results (GB, FB, LD rates).
Friend:why do you love ripping on Jeff Francoeur?
Me: Because hes an idiot, who posseses huge physical tools but will not, nay, cannot, harness those tools into any sembelance of usable baseball skill and refuses to make any effort to do so. Hes like a tragic greek figure. doomed by his own stupidity and 80 raw power.
xFIP is more or less doing this.
It adjusts HR allowed so that HR/FB rate is constant across all pitchers. So it’s (sort of) using using GB% rather than HR/9.
SIERA uses GB% explicitly. It does not seem to be a meaningful improvement over xFIP in terms of prediction, but it’s a good stat that has some nice diagnostic uses that xFIP doesn’t.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
by JinAZ on Aug 6, 2010 4:54 PM EDT up reply actions 1 recs
so explain xFIP
he seems to have been talking about FIP which doesnt take that into account. all I’m trying to say is, while the tone of the title was confrontational and it seemed trollish, there were some decent points to be addressed here, and we should read what he has to say before we go and call people idiots
Friend:why do you love ripping on Jeff Francoeur?
Me: Because hes an idiot, who posseses huge physical tools but will not, nay, cannot, harness those tools into any sembelance of usable baseball skill and refuses to make any effort to do so. Hes like a tragic greek figure. doomed by his own stupidity and 80 raw power.
by harendaman365 on Aug 7, 2010 11:16 PM EDT up reply actions
there were some decent points to be addressed here, and we should read what he has to say before we go and call people idiots
Just a theory, but a reason for the reaction might be that everyone already understands these points and people have been putting a lot of thought and energy into identifying how big of a problem they are and the best method to solve them. The results have come back to say FIP/xFIP work well, despite these concerns – because these issues just aren’t that big of a deal. The research is out there. People have addressed these concerns. If someone doesn’t bother to read up on the knowledge that’s available before coming to a site like BtB (which is basically SBN’s saber site, in my eyes) and posting a fanshot titled “FIP is a Garbage Statistic”…..well, are you really expecting a friendly reaction? I don’t think people need to address his concerns here at all, he’s obviously a troll, and his concerns have been, are being, and will continue to be addressed by the saber/blogging community for the foreseeable future. He’s welcome to read them.
by Missing Barry on Aug 9, 2010 5:21 PM EDT up reply actions

by 



















