FanGraphs Baseball


RSS feed for comments on this post.

  1. Well, Bill, if “respectable baseball writers” link to you when they’re high, I’ll take this analysis as gospel.

    Comment by Delino DeShields — March 14, 2012 @ 12:10 pm

  2. Pretty sure this would be difficult to prove, but I think players like Fielder (big power guys who hit long HRs) are less affected by changing parks, as most of their HRs would be out almost anywhere anyway. For a player like, say, Ichiro (trivial-to-modest power guys who don’t hit bombs), moving from Safeco to Yankee Stadium should increase his HR total by a larger percent, since he would be hitting “just enoughs” in Yankee Stadium that would be outs in Safeco. Not sure if this was considered, but it seems the HR/PA chart bears that out.

    Comment by DD — March 14, 2012 @ 12:58 pm

  3. If the change of park does not have much effect in general, that brings the park factors themselves into question. Are they possibly wrong or at least overrated?

    Comment by Baltar — March 14, 2012 @ 1:33 pm

  4. Also, could we do this same analysis comparing the players’ home/road splits and the change in them when changing parks? For example, if Carlos Gonzalez changed home parks to a neutral environment, would his home/road split variance shrink?

    Comment by DD — March 14, 2012 @ 1:43 pm

  5. Something here doesn’t make sense to me. Power hitters are more affected by the park change when moving to a less hitter friendly park, but less affected by the park change when moving to a more hitter friendly park? Shouldn’t the changes be symmetrical (as in the park’s attributes either always affect power hitters more, or always affect power hitters less)?

    Comment by Jono411 — March 14, 2012 @ 1:44 pm

  6. I first read this as the impact of hitters who physically change a park. So like how Tampa changed the LF wall for Crawford and such. Probably not enough data on this, but would be interesting to see if this ends up helping a team or not.

    Comment by Julian — March 14, 2012 @ 1:54 pm

  7. What was the sample size here? Have we demonstrated that the expected random variance in HRs as measured by % of total is the same in low-power and high-power hitters? I’d want to see an enormous sample of <10 HR/600PA guys getting 5x the HR boost than park factors would lead you to expect. Adding lefty/righty splits (which fractures the sample) and using home/away data for all players, instead of just those who switch teams (which exponentially increases the sample) would, I think, make this a stronger study.

    Comment by byron — March 14, 2012 @ 2:16 pm

  8. Really cool stuff.

    One thing I’m concerned about here are the lines where hitters are split in >.145 ISO in year 1 and <.145 ISO in year one. We'd expect regression to the mean to play a role when we split hitters like this. The first group would see lower power numbers than expected (based on the park factor change) and the second group would see higher power numbers than expected based on park factors. In other words, I think this:

    [quote]On average, above-average power hitters in the sample moved to parks that were on average 28.6% more home-run friendly, but only increased their HR/PA by 3.6% — an almost 8:1 ratio.[/quote]

    can be explained in part by regression to the mean.

    Anyway, great study. So, hitters get a *bigger* HR swing than expected (less than 2:1 ratio) but a much smaller *wOBA* swing than expected (much larger than 2:1) ratio, right? This implies that non-HR components of wOBA must be *much* less affected by park changes than expected.

    I'm going to have to check the Seamheads site to see if their ballpark factors are regressed.

    Comment by J. Cross — March 14, 2012 @ 2:18 pm

  9. Yes. Park factors as we know them simply are not useful. Fangraphs has not come around on this yet. It will be better baseball analysis site if and when it finally does.

    Comment by Jack Nugent — March 14, 2012 @ 2:48 pm

  10. I think J. Cross’ post a couple down explains this – it’s because of regression to the mean.

    Assuming the positive change and negative change groups have the same samples size, we’re seeing an average -10% change in HR/FB rate for the high power group. And therefore the effect of the ballpark change is really around +/-13% with approximately a 25% change in park factors, which is the 2 to 1 ratio you’d expect.

    For the low power group, we have an average +25% change in HR/FB rate, and the effect of the ballpark change is really around +/-35%, with approximately 25% park factor changes. So in this case, we have a 0.7:1 ratio.

    Comment by Jono411 — March 14, 2012 @ 2:50 pm

  11. Does it matter that home runs are, famously, a “true outcome”? Overall park factors are quite complex, involving amount of foul territory, the hardness of the infield and height of the grass, speed of basepaths, walls to hit doubles off of, etc. All these affect team scoring, but they aren’t going to track consistently onto individual performance

    Comment by Mr Punch — March 14, 2012 @ 3:24 pm

  12. “I’m going to have to check the Seamheads site to see if their ballpark factors are regressed.”

    The single-year factors are unregressed.

    The 3-year factors are regressed to the long-term, lifetime park factor for that park. So for example Coors averaged for 3 years (1 year before, current year, and 1 year after) then regressed 25% to the Coors specific lifetime factor. We won’t claim they’re perfect factors, but we believe regressing to a park specific number is more accurate than regressing all parks to average.

    Comment by KJOK — March 14, 2012 @ 3:43 pm

  13. Mind providing a link to an article that makes an argument for this point. Bare assertions are simply not useful.

    Comment by LTG — March 14, 2012 @ 5:59 pm

  14. How does a massive outlier like Adam Dunn affect these data? He went from a bad hitters park to a great one, yet cratered. Does that appreciably affect the numbers?

    Comment by MikeS — March 14, 2012 @ 6:18 pm

  15. I would think that big power guys who hit long home runs would be pretty well represented by guys with ISOs over .145

    Comment by Ben Hall — March 14, 2012 @ 8:07 pm

  16. Anecdotally, there is a massive home/away split in Dustin Pedroia’s wRC+, a split that shouldn’t exist over such a large sample assuming park factors are correct.

    Comment by YanksFanInBeantown — March 14, 2012 @ 8:37 pm

  17. Another point of interest for me is how do Power hitters affect the park factors?

    Are the park factors completely independent of player performance? For instance, does Miller Park’s park factor go down without Prince in the Brewers’ line-up for the season?

    Comment by RC — March 14, 2012 @ 9:38 pm

  18. I think the biggest issue for Fielder will be league pitching.

    Since players get 1/2 of their PA on the road, you might look at weighted road park factors as well.

    Comment by pft — March 14, 2012 @ 9:42 pm

  19. Is there any reason not to expand this study to include the differences experienced by all players in the different environments they play in . . . also known as computing park factors?

    Comment by byron — March 15, 2012 @ 12:10 am

  20. Several issues, most of them already touched upon:

    One, as Jared (Cross) points out, the regression towards the mean for the high and low ISO players is a critical factor which was overlooked by the author. As Jared points out, that would partly or completely explain the assymetrical park effects. All of the high ISO players will have their ISO go down in year 2 regardless of whether they changed parks or not or moved to a hitters or pitchers park. Same thing in reverse for the low ISO guys.

    Two, the park factors listed on Seamheads and used by Bill are by no means very accurate for several reasons. One, with inter-league games and the imbalanced schedule, computing park factors in the traditional manner, home and road team stats in one park divided by home and road team stats in all other parks, is not nearly good enough anymore.

    Three, the way they do the regression for the 3-year PF’s (I was not aware that they were regressed) is statistically unsound (you do not regress a sample mean toward a larger sample mean, you regress it toward a population mean, where the population is a large group to which the sample element belongs). There is some justification for doing it the way they do it, but not much. And 25% is not nearly enough for run factors, since as someone already pointed out, some aspects of PF’s, other than the HR’s, are very variable (i.e., they have a narrow range in “true” values and of course a large random variance component in any 1 or even a few years).

    Four, you cannot compare wOBA to a run factor, straight up! You have to convert wOBA into runs or square it as a short cut. For example, if a true PF were 2.0, i.e., it doubled run scoring, do you think that it would double wOBA also? If league wOBA were .330 and a team scored 5 runs per game, and then their wOBA were .660 in that same park, how many runs do you think they would score? Like 16 runs! In order to score 10 runs, you would need a wOBA of around .475. which is 1.44, Sqaure that and guess what? You get around 2! So Bill needs to square the wOBA in year 1 and year 2 before he computes the ratio!

    Five, and finally, park factors cannot be treated the same across leagues! They are strictly league dependent and league relative. There is no particular reason to think that if a player with a home park factor of 1.00 moves to the other league and to a home park with a park factor of 1.02, that is offensive stats will go up. First of all, the two PF’s are not equivalent. The entire set of parks in one league may be more or less hitter friendly than the entire set of parks in the other league. They likely are (by a little). Two, since the pitching is different (and we think that the AL pitching is much better), a player going from one league to the other may see his offensive stats go up or down because the quality of the pitching may change. Three, when as player goes from one league to the other, he likely has an adjustment period where his stats are not as good because he is not as familiar with the pitching.

    Comment by MGL — March 15, 2012 @ 1:12 am

  21. the pitching in the AL Central sans Detroit is craptastic….his hardest matchups could be against who? Danks? Ubaldo?

    78 games against staffs that pretty much suck….very weak on the high end talent and lots of #3 and 4 slot starters….not much top end talent in the bullys either….

    I like Prince for 42 dingers and 115 rbi and about 100 walks again

    Comment by Tim — March 15, 2012 @ 1:50 am

  22. Milwaukee brought in their right field wall for Geoff Jenkins and just in time for Fielder.

    Comment by Dealer A — March 15, 2012 @ 2:07 am

  23. Prince hits some of the longest HRs in all of baseball — which others have mentioned — so I don’t really expect the change in venue that to affect him that much. HitTrackerOnline backs this up to some extent: only 2 of his HRs wouldn’t have been gone in at least half the ballparks in baseball, and only one of those was at home (and that one still flew almost 400 feet). I would be far more concerned about age (and weight) related decline/injury. If he has a large decline in HRs, this will probably be the reason, not park effects.

    Comment by Brandon T — March 15, 2012 @ 8:31 am

  24. You’ll have to provide a reason for you to think he has a “large” sample.

    Comment by Tangotiger — March 15, 2012 @ 9:28 am

  25. Tried to respond yesterday, but for some reason FG kept eating my comment. Here’s why park factors are useless:

    Comment by Jack Nugent — March 15, 2012 @ 1:40 pm

  26. You are forgetting the other 72 games against the AL East and AL West.

    Also, the NL Central sans Milwaukees was not much better than the AL Central will be in 2012. Indians and White Sox look to have decent staffs this year.

    3 yr HR PF (LHB)

    DET-89 (Milwaukee-116)

    (Avg sans Det-93, NL Central sans Milwaukee-102)

    AL West has neutral parks (avg HR LHB PF – 100) and tougher pitching. AL East has tougher pitching and smaller parks (Avg HR LHB PF 108).

    Add to this, Prince has not seen much of most AL pitchers, and that’s usually an advantage for the pitchers, so he should have an adjustment period. Also, players signing large contracts tend to press more to justify their contract and that can hurt performance (not always).

    I think Prince has a tough 1st year and will be lucky to hit 30 HR and get to a 900 OPS this year .

    Comment by pft — March 15, 2012 @ 10:12 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.

0.264 Powered by WordPress