Park Factors Added to FanGraphs Guts!

In the FanGraphs Guts! page, you’ll now find a bunch of component park factors, included park factors by handedness.

All park factors are calculated using this 5 year regressed version.




Print This Post

David Appelman is the creator of FanGraphs.

26 Responses to “Park Factors Added to FanGraphs Guts!”

You can follow any responses to this entry through the RSS 2.0 feed.
Click here to view comments in a non-threaded output.
  1. Gheto Sparetto says:

    Great effin stuff

    Vote -1 Vote +1

  2. Ivan Grushenko says:

    How is Oakland’s park factor for 3B 100 overall, but 89 and 88 for LHB and RHB?

    Vote -1 Vote +1

  3. chuckb says:

    Will you marry me, David?

    Vote -1 Vote +1

    • chuckb says:

      Sorry, in my haste I forgot to ask how to access the “guts” page from the homepage. I didn’t see it instantly in the glossary.

      Vote -1 Vote +1

  4. Slartibartfast says:

    I’m finding the explanation of the PF calculations a bit lacking in the link you provided. The author is scatter-brained in an attempt to be thorough, and following his train of thought is headache. Is there a FG authored or condensed version? The math explained/derivation should be able to be expressed in a few paragraphs without all that clutter.

    Vote -1 Vote +1

  5. Jack Nugent says:

    Please, please, please read this before you lose your shit over how awesome it is that Fangraphs is listing Park Factors that aren’t worth a…

    Well, decide for yourselves.

    http://highboskage.com/stat-corrections.shtml

    Vote -1 Vote +1

    • Please read how these park factors are computed.

      Vote -1 Vote +1

      • Tyler says:

        Where is the park factor data on the site?

        Vote -1 Vote +1

      • Slartibartfast says:

        David, if you’re solely relying on that linked article to provide the explanation of your PF, I think you’re doing a FG a disservice. It’s not that the article isn’t enough to decipher much of the methodology if you try hard enough, it’s just that the article is far from user friendly for anyone of any level of baseball knowledge. It’s just bad.

        Vote -1 Vote +1

      • Jack Nugent says:

        More importantly, it doesn’t appear to address at least one of the biggest hurdles preventing park factors from being useful.

        How many ballparks have been in existence for fewer than five years? Off the top of my head I can think of several. And what to make of parks experiencing major structural changes (CitiField).

        Regression is a great tool. I don’t think it sufficiently cures park factors of their woes.

        Vote -1 Vote +1

      • Jack Nugent says:

        From the link you provided:

        “…These weights seem arbitrary, but maybe he (MGL) had a good study to base them on. Anyway, I think their reasonable and that’s why I use them.”

        Vote -1 Vote +1

      • Jack Nugent says:

        Sounds like rock solid methodology to me…

        Vote -1 Vote +1

  6. Jonathan says:

    D-d-d-do you have it?

    Vote -1 Vote +1

  7. Paul Clarke says:

    David,

    Speaking of park factors, I posted a comment on the “Change of Scenery” thread about how the wrong park factor seems to be applied to home/road wRC+ splits. Could you take a look and tell me whether or not I’m talking rubbish? I’ve pasted the original comment below. Thanks.

    —————
    Iannetta’s offense was mostly isolated to Coors field as his home/road wRC+ splits sat around 120 and 80, respectively.

    I strongly suspect that this is because home/road wRC+ splits are calculated using the wrong park factor. wRC+ is basically (wRC/avg wRC) plus a park correction, and we can calculate avg wRC as (wRC – wRAA), so a non-park-corrected wRC+ (call it wRC+’) is wRC/(wRC – wRAA). Iannetta’s splits are:

    home: wRC = 140.6, wRAA = 34.9
    road: wRC = 102.2, wRAA = -4.1

    So:

    home wRC+’ = 140.6/(140.6-34.9) = 133
    road wRC+’ = 102.2/(102.2 + 4.1) = 96

    His actual wRC+ splits are 121 and 83, so the park correction is subtracting about 12 or 13 points in each case. That implies that his average road park has the same park factor as Coors, which is clearly wrong. The same is true of his overall career numbers (wRC+’ 114, actual wRC 102), so it looks like his average park factor (roughly 50% Coors, 50% road parks) is being applied to his home/road splits.

    Vote -1 Vote +1

  8. Keith says:

    I’ve seen PF presented a couple ways accross the net, so please forgive the ignorance here…

    Are these factors for the actual parks, or for reference agains FY stats of the players who play in those parks? It’s essentially the difference between, for example a 110 HR PF meaning the park itself inflates HRs by 10% or that players on that team should have 10% inflated HR totals because their park inflates HRs by 20% (but they play half their games there).

    Vote -1 Vote +1

  9. Nadingo says:

    Is there a link to Fangraphs Guts on the home page? I haven’t been able to find it.

    Vote -1 Vote +1

  10. Greg says:

    Last year I saw a spray chart of home runs hit at Yankee Stadium, and they superimposed Citi Field on the chart. Only 50-52% of the home runs hit at New Yankee Stadium since it has opened would have been home runs at Citi Field, yet park factors would suggest that the gap is not nearly that large. Derek Jeter would have lost over 80% of his home runs.

    Unfortunately, I just don’t think park factors are nearly accurate enough yet. Adrian Beltre did not suddenly become a better hitter after leaving Safeco for more friendly hitters parks, he just left a park that played directly against his strengths.

    Vote -1 Vote +1

  11. Herbstr8t says:

    HR park factors are only helpful if they are done by handedness, as included on StatCorner for example. The split for Fenway is 79/94 for LHB/RHB. That’s a hugely important difference that’s not currently captured on the GF guts page.

    Vote -1 Vote +1

  12. docmex says:

    I’ll have to apologize in advance for my ignorance but someone explain why Ks, IFF, LD, GB, etc would be impacted by a ballpark? The differing dimensions obviously have an impact on things like HRs (short fences), and XBH (large OF area) but why would a ballpark have any impact on how often a hitter strikesout? or walks? or hits a LD or IFF? Those all seem dependent on the pitchers a hitter faces in a given park rather than by any features of the ballpark itself. This seems more likely as well since most of the factors in these categories fall within a very narrow range and few if any seem to have any statistiscal difference at all if you accept random fluctuations.

    Vote -1 Vote +1

    • Xeifrank says:

      Different foul ball territory sizes effects SO and BB. Smaller foul ball territory means less pop ups caught and at bats extended. Extended at bats lead to more SO and BB. Shorter at bats due to foul balls caught, of course lead to less SO and BB as the at bat is over.

      To a lesser degree “could” be hitter background and mound.

      Vote -1 Vote +1

  13. Tom says:

    David….. what park factors are used in the pitcher adjustments (things like WAR or FIP-)

    My understanding is that the run factor is used but what happens for parks that have very different HR and run park factors. Examples:

    Fenway which has a high run factor but a low HR factor.
    Yankee stadium – slightly above average run factor but a significant HR factor
    San Francisco – low run factor, but very low HR factor.

    If the pitchers are having the HR’s impacted differently (which obviously impacts FIP and WAR), but you are using run factors for the WAR, FIP- adjustment does it lead to issues in parks which have a signficant difference between run and HR factors?

    Vote -1 Vote +1

  14. JD says:

    I’d like to see park factor splits. I’d love to see the park factors for an April/May night game at Petco

    Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

*