FanGraphs Fantasy Baseball


RSS feed for comments on this post.

  1. Well…FIP definitely helped predict Ricky Nolasco’s turnaround. Not sure what his xFIP was….

    Comment by R M — July 16, 2009 @ 2:03 pm

  2. Rich Harden comes to mind, given that he’s given up a bunch of HRs. You should just sort by HR/Fb to find the differences between xFIP and FIP, since HR rate is the only difference in the computations.

    I also noticed Porcello, Saunders, Galarraga, etc. There’s probably others, but I didnt feel like going through the entire list.

    Comment by Steve — July 16, 2009 @ 3:05 pm

  3. I think it is good to pay attention to if you’re playing in a league where no one else does. Then you can unload pitchers with sterling ERAs, but high strand rates/low BABIPs, for other pitchers or hitters with more “true talent”. Again, it doesn’t always work out to your advantage, but it can.

    And in the case of Nolasco, someone in my league dropped him once his ERA hit 7.00; I got excited and I checked his FIP, then picked him up immediately, even though he was in the minors. Again, this is in a league where people make waiver claims based on pitching wins, Yahoo! rank, and batting average, so maybe it’s meaningless.

    Comment by Drew — July 16, 2009 @ 3:09 pm

  4. It is curious that there are so many legitimately good pitchers on this list. It makes me wonder if the entire premise is wrong. Maybe better pitchers can legitimately produce a lower HR/FB rate. Is there some documentation of immutability of the HR/FB rate?

    Comment by baseballfan — July 16, 2009 @ 3:27 pm

  5. Like Drew said above, using FIP or xFIP is a big advantage in leagues where no one else does. Unfortunately, the guys in the leagues I play know all about both. Thus, some with good traditional stats, like Mark Buehrle (9 wins, 3.66 ERA, 1.19 WHIP) is not worth much in a trade because of his 4.63 FIP and 4.49 xFIP.

    This year, I’ve been using the midpoint between FIP and xFIP to make pitching roster decisions, mainly because I’m not sure which is better. I think a normalized HR rate makes sense, so I lean towards xFIP as being the better measure.

    I’ll be very interested in how your study turns out.

    Comment by BobbyRoberto — July 16, 2009 @ 3:34 pm

  6. Thanks Steve.

    I added Harden and Porcello to the list but Saunders and Galarraga did not meet the 0.50 qualification.

    Comment by Brian Joura — July 16, 2009 @ 3:36 pm

  7. I agree. I don’t have too many FanGraphs-types in my league, so I use FIP and xFIP to my advantage and have been very successful targeting and moving my pitching to maximize player value in my league.

    I can see how it’d be difficult in other leagues though.

    One sidenote — can xFIP be added the the leaders and player pages? I think this would be very helpful.


    Comment by Rob — July 16, 2009 @ 4:11 pm

  8. Hey R.M.,
    That was my article Brian cited, so I thought I’d jump in here. I thought your comment was a good one for all fantasy players to read, so I posted it in a quick article at THT (click my name to go to it). Here’s how I addressed your comment:

    We must remember that FIP is not so utterly useless that it will be incorrect in every scenario. In scenarios where the pitcher has a lucky or unlucky BABIP or LOB% (Nolasco’s BABIP was over .400 at one point), FIP will be able to predict the general direction the pitcher’s ERA should move as long as the HR/FB isn’t too far away from league average.

    While we’ll know that Nolasco isn’t a 6.00 ERA pitcher, it is important to make a distinction over whether his ERA should be 4.50 or 4.00 or 3.50. Even the difference between a 4.25 and 4.00 ERA is the difference between ‘solid starter’ and ‘waiver wire material’ in many leagues. FIP is ill-equipped to make this distinction.

    We can’t allow anecdotal evidence to rule our decision making. While FIP may have worked in Nolasco’s case given a very rough objective, the numbers tell us that a stat like xFIP or LIPS will be more accurate, for more pitchers.

    Comment by Derek Carty — July 16, 2009 @ 4:47 pm

  9. If we can’t predict spikes or dips in HR/FB, isn’t it better to use xFIP all the time with the expectation that regression to the mean is the most likely outcome as the sample size increases? While a pitcher may get lucky over the course of a season, from a statistical standpoint we can’t assume that their luck will continue, as FIP does. The obvious example to me is rolling a six sided die – even if we roll a 1 25 times out of the first 100 rolls, that doesn’t mean we should alter our expectations over the next 100 rolls and assume that we should roll a 1 25 times again. It makes sense that HR/FB should follow the same rules: even if a pitcher has an abnormally high or low HR/FB over the first half, the best we can due is assume that their hr rate will regress to the mean because they have no control over the outcome of fly balls put into play.

    Comment by EricG — July 16, 2009 @ 4:54 pm

  10. Bobby, I’d strongly implore you to stick to xFIP entirely. I just posted an article at THT (click my name to read it) talking some more about this, but here’s a recap:

    Looking at yearly pairs dating back to 2004 (i.e. 2004/2005, 2005/2006, etc.), there were 63 pitcher seasons where the pitcher posted a HR/FB that strayed 4% of more from league average in Year 1. In Year 2, just 5 of those 63 pitchers failed to regress in the direction of league average (7.9%). That’s an incredibly small number, especially when you consider that Chien-Ming Wang (who may be one of the rare exceptions we allow for) and Brett Myers (who almost certainly is one of those rare exceptions) accounted for 2 of those 5 seasons. Exclude them, and the percentage becomes 4.8%.

    This is a very crude study, but hopefully it reestablishes my point. David Gassko did some much more thorough work on HR/FB in the THT Annual 2007 Short version is that for pitchers with 350+ BFP, David found that the previous season’s HR/FB explains just 3% of the variance of the following season’s HR/FB.

    Comment by Derek Carty — July 16, 2009 @ 5:04 pm

  11. HR/FB is a factor mostly of the park as well as what hitters the batter faces. The NL West is not a very powerful league and has a lot of big parks. Most of the starters in that league have lowered HR/FB, but it is not a factor they are controlling.

    Comment by Troy Patterson — July 16, 2009 @ 7:19 pm

  12. Can we also throw in a normalized BABIP and Strand Rate? I have no clue how hard that would be.

    Comment by Josh — July 16, 2009 @ 7:28 pm

  13. I strongly suggest this study at THT by Colin Wyers. He ran a split comparison using FIP, xFIP, tRA and ERA. He tried to see which stat when used from half a pitchers games would most accurately predict the other half. You can look at the data here..

    xFIP is the most accurate of these four and ERA is the worst as we would expect. FIP and xFIP (as well as tRA) are not so horrible you couldn’t argue against using any of them. I like to look at FIP and xFIP as well as understanding the limitations of them. You can improve your use and value of FIP and xFIP if you look at HR/FB and know when it should regress and when it might not…Matt Cain

    Comment by Troy Patterson — July 16, 2009 @ 7:29 pm

  14. FIP and xFIP account for a normalized BABIP and LOB%. That is the central goal of FIP.

    Comment by Troy Patterson — July 16, 2009 @ 7:30 pm

  15. When I use FIP, I don’t tend to just take the number as granted. I check out their three year HR/FB rate and BABIP/LD%, and make a healthy guess on how they will perform on these two rates going on with their FIP. I don’t use xFIP since I believe that some pitcher has lower HR/FB rate and LD%(thus lower BABIP)

    Comment by Kampfer — July 16, 2009 @ 8:36 pm

  16. Why doesn’t fangraphs include xFIP?

    Comment by baseballfan — July 16, 2009 @ 9:53 pm

  17. How to calculate FIP and xFIP?

    Comment by Tiago — July 16, 2009 @ 11:11 pm

  18. Personally I use both tRA and FIP in combination, although I”m not really sure yet how well that is working.

    Doing great in pitching in two leagues, and awfully in another.

    Comment by Andy S — July 17, 2009 @ 12:22 am

  19. Why are you using this years data? Couldn’t you just use last years half way point data and compare that to the second half data of the season? I’m confused as to why you are choosing a sample that is forcing you to wait a few months.

    Comment by Davidceisen — July 17, 2009 @ 10:37 am

  20. Sorry if you’ve already addressed this elsewhere, but how well does xFIP deal with park effects? When Jamie Moyer, Joe Blanton, and Cole Hamels all have HR/FB% above 13%, is that something that one should expect will regress, or is that more a factor of their pitching environment?

    Comment by NadavT — July 17, 2009 @ 10:52 am

  21. NadavT,
    xFIP doesn’t account for park effects. To the best of my knowledge, LIPS ERA is the only ERA estimator that does. However, the effects of Citizen’s Bank (or any park for that matter) shouldn’t lead to a HR/FB above 13%.

    If we approximate league average to be 10% HR/FB, for the most extreme HR park in baseball (U.S. Cellular Field – PF = 1.26), we’d expect a pitcher to post a 12.6% HR/FB while pitching at home. If we assume a league average mix of road parks though (and an even playing time split), his overall HR/FB would only be 11.3%.

    So while Citizen’s Bank inflates HR/FB by 16.4%, this won’t lead to any outrageous ratios in isolation. I’d expect those guys to see some regression.

    Comment by Derek Carty — July 17, 2009 @ 11:50 am

  22. I do not have access to the necessary data from mid-season in previous years.

    Comment by Brian Joura — July 17, 2009 @ 12:03 pm

  23. Thanks for clearing that up. I think, though, that it’s important to note that in the small samples that you often deal with in fantasy baseball, it’s not always safe to assume a league average mix of road parks, or even an even split of home/road starts. So far this season, for example, Hamels has had 10 of his 17 starts at home, and his away games have included one start apiece at Yankee Stadium, Great American Ballpark, and Coors, so I would think that we wouldn’t expect quite as much regression if he continues to pitch in similar environments.

    Comment by NadavT — July 17, 2009 @ 12:33 pm

  24. For those who want more information on all things FIP, Troy Patterson (who commented earlier in this thread) has some info up at his site

    Comment by Brian Joura — July 17, 2009 @ 2:31 pm

  25. I don’t know about Jimenez’s xFIP, but his ERA his 3.81 and according to Fangraphs, his FIP is 3.27, making a 0.50 difference.

    Comment by bballrox4717 — July 17, 2009 @ 2:51 pm

  26. Also, De la Rosa and Hammel have even larger differences than Jimenez.

    Comment by bballrox4717 — July 17, 2009 @ 2:54 pm

  27. Hi bballrox thanks for reading and commenting!

    The test is between FIP and xFIP not ERA, since it is known that FIP does better than ERA in predicting future ERA. I am only looking for a 50-point difference between FIP and xFIP.

    Comment by Brian Joura — July 17, 2009 @ 2:57 pm

  28. You’re right Nadav, but the problem is that Hamels won’t experience that same mix of road parks going forward, so assuming he will is flawed thinking. This kind of thing is completely out of his control, so unless we want to go park by park and at-bat by at-bat and make adjustments (which I do in the off-season), the simple way is to simply assume that he’ll have neutral luck and a neutral mix of parks going forward.

    Comment by Derek Carty — July 17, 2009 @ 10:02 pm

  29. Brian (or anyone else who might have an idea),
    What do you think of QERA, which is sometimes used at BP? It uses K-rate, BB-rate, and GB-rate. I calculated the QERA for pitchers on my fantasy teams and it comes out fairly close to xFIP.

    Comment by BobbyRoberto — July 18, 2009 @ 1:11 pm

  30. BobbyRoberto,
    My tests have shown that QERA is a pretty good one, which it logically should be since K, BB, and GB are really what we need to be focusing on. The ERA estimators we really want to avoid are FIP and ERC.

    Comment by Derek Carty — July 18, 2009 @ 7:30 pm

  31. derek,

    Where do you think tra and tra* fit into the picture? which of the two is better for predicting future performance? thanks.

    Comment by labe — July 18, 2009 @ 11:03 pm

  32. FIP
    Fielding Independent Pitching, a measure of all those things for which a pitcher is specifically responsible. The formula is (HR*13+(BB+HBP-IBB)*3-K*2)/IP, plus a league-specific factor (usually around 3.2) to round out the number to an equivalent ERA number. FIP helps you understand how well a pitcher pitched, regardless of how well his fielders fielded. FIP was invented by Tangotiger.

    Expected Fielding Independent Pitching. This is an experimental stat that adjusts FIP and “normalizes” the home run component. Research has shown that home runs allowed are pretty much a function of flyballs allowed and home park, so xFIP is based on the average number of home runs allowed per outfield fly. Theoretically, this should be a better predicter of a pitcher’s future ERA.

    xFIP multiplies the number of outfield flys by 11% to get the predicted number of HRs (although it should probably be closer to 10% and for fantasy purposes should be adjusted by 50% toward the player’s home park).

    Comment by Toffer Peak — July 19, 2009 @ 1:52 pm

  33. labe,
    tRA is one I would probably avoid as well. I imagine it’s better than ERC and probably FIP, but worse than LIPS, xFIP, QERA, etc. tRA uses HR and LD, both of which are mostly out of a pitcher’s control, so it has much the same problem as FIP.

    tRA*, however, regresses each stat to the mean to account for the fact that HR/LD/etc aren’t heavily controlled by the pitcher. While I prefer LIPS, tRA* doesn’t seem to have any of the glaring flaws that FIP and ERC do. I haven’t run any tests on tRA*, but I imagine it’s at least in the same class as xFIP and QERA based on its methodology.

    In summary, if you’re going to use one of the two, definitely go with tRA*, although xFIP and QERA are also great freely available options.

    Comment by Derek Carty — July 19, 2009 @ 2:24 pm

  34. I guess we’ll just have to disagree on this one. If you’re taking the time to project a pitcher’s performance for the rest of the season, I don’t think it’s sufficient to go with the “simplest way” by assuming that his mix of ballparks will be HR-neutral. Clearly, you can’t pin down every ballpark he’s going to start in for the rest of the season, but you know that his home games will be HR-friendly and you can make guesses based on the remainder of the Phillies’ schedule. I’m not suggesting that one can apply a mathematical adjustment to FIP or xFIP based on this, but I maintain that it makes sense to keep in mind that xFIP might be biased downward in projecting his performance for the rest of the season.

    Comment by NadavT — July 20, 2009 @ 4:11 pm

  35. Nadav,
    “assuming that his mix of ballparks will be HR-neutral” and “assuming that his mix of *ROAD* ballparks will be HR-neutral” are two completely different things. We should absolutely account for home park separately, but unless we want to get super in-depth mid-season (like I do with CAPS during the off-season), it’s impossible to put an exact number on a pitcher’s future road ballparks. You admit this, so I agree that it’s fine to make mental adjustments if you see that a player’s upcoming games are in extreme parks. My point remains, though, that xFIP is better than FIP and that assuming a neutral mix of road ballparks is fine unless you want to take the time to dig through schedules.

    Comment by Derek Carty — July 20, 2009 @ 11:52 pm

  36. what were the results???

    Comment by David MVP Eckstein — February 9, 2010 @ 9:04 pm

Leave a comment

Line and paragraph breaks automatic, e-mail address never displayed, HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Close this window.

0.374 Powered by WordPress