An In-Depth Look at Dodger Stadium’s Effects

Each ballpark has different effects on each player. Since each team is composed of different players and since a team can choose players who will benefit from playing in the ballpark the team calls home, there is an innate potential of selection bias in park factors. Furthermore, the effects a ballpark has are interrelated. A long fly ball can be a home run in one park and an out in another; a long fly ball hit with less hang time could be a home run in the first park and a double in the second.

On top of that, even 81 full games of data are far from an optimal sample size. 243 full games of data is a sub-optimal sample size. In addition, new parks are often introduced into the league and different teams visit different parks with differing frequencies and at different times season to season, which is of significance since weather plays a substantial role in how many runs are scored in any given game. As such, measuring park factors gives us a rough idea of the park’s value to the run environment, but doesn’t quite answer the question of the park’s true talent level, so to speak. It is with these caveats in mind that I now embark on a journey to elucidate the impact of Dodger Stadium on the baseball games it hosts.

You’re surely familiar that Dodger Stadium is typically regarded as an extreme pitcher’s park. Its one-year park factors, based on runs per game at home divided by runs per game on the road, have been .908, .868, and .825. That sample is pretty small, so suggesting that Dodger Stadium has become less severe as a pitcher’s park over the last three seasons is premature and likely inaccurate. It is, however, a sample that suffices in making the point that Dodger Stadium favors defense and hurts offense.

The Dodgers have been a pretty good team recently. Over the past three seasons, they’ve gone 141-102 at home and 129-114 on the road. That has a significant impact on the park factor that one computes for them. If you use the typical method based on runs scored and allowed, home and away, you make Dodger Stadium look more like a pitcher’s park. Why? Because the Dodgers are making more of their road games last nine full innings and making fewer of their home games last nine full innings, and fewer innings means fewer runs. If Dodger Stadium were completely neutral, any 2002-2004 park factor based on counting stats will make it appear to be more of a pitcher’s park. A sliver of Dodger Stadium’s reputation as a pitcher’s park can be attributed to Los Angeles only having ten seasons under .500 in 43 years at Chavez Ravine.

The effect of that particular bias has been significant. Using runs scored, Dodger Stadium has a park factor of .86 for the Dodgers’ offense and .87 for the Dodgers’ defense. Prorate to runs per out, however, and the park factor has been .92 for the Dodgers’ offense and .83 for the defense. The difference between those two figures is what we typically call home field advantage.

To get a better idea of what Dodger Stadium’s impact is, we need to look into how it suppresses run scoring. And here it’s important to consider that the same issue that makes Dodger Stadium appear to hurt runs slightly more than it does in park factors applies to other types of park factors. The park factors tracked last season by ESPN, for instance, only measure each offensive event by the ratio of counting stats, not by the ratio of the event’s frequency. Instead of measuring home runs per at bat, it’s measuring home runs per game.

If you look at that data, you’ll see that Dodger Stadium had an extreme park factor for walks last season. If there is a disparity in walks, we would expect a disparity in hits and home runs; fewer walks mean more opportunities for all the other offensive events. As such, it is more illuminative to consider the frequency of each offensive event in the context in which it occurs. That is to say, we should break things down into walks per plate appearance, strike outs per at bat, home run per at bat minus strikeout, hits per ball in play, and so forth, an approach familiarized by Voros McCracken and Tangotiger. This will then isolate the effect that Dodger Stadium has on each kind of event.

We can do even better than that, though, with batted ball data. To capture each event’s frequency, I’ve broken down each event into this chain: IBB, SH, Foul Out, HBP, BB, K, Line Drive, Fly Ball, Ground Ball. To calculate the frequency of a strikeout, K is divided by (K+LD+FB+GB) for $K. For $LD, it’s LD/(LD+FB+GB). I’m not completely confident that this is the best way to divide things up, but if the data is used correctly that shouldn’t be an issue. Within each type of batted ball, the chain is HR, Outs, Extra Base Hits, and Triples, with a separate reached on error calculation ($ROE = ROE/(ROE+Outs)). Foul outs have been removed from the fly ball totals.

Here are the observed park factors for 2002-2004, based on STATS Inc. data, and excluding interleague play (the last column, “r/o”, is the park factor for Base Runs (David Smyth’s run estimator) per out for each type of batted ball):

            $foul    $bb     $k    $ld    $fb    $gb
Offensive   1.261  1.011  1.002  1.059  0.963  1.027
Defensive   1.288  0.892  1.086  1.051  0.908  1.070
Total       1.274  0.951  1.044  1.055  0.936  1.048

GB            out    $hr     $h    $xb    $3b   $roe    r/o
Offensive   1.018      -  0.933  0.941  0.411  1.048  0.869
Defensive   1.040      -  0.867  0.779  div/0  0.970  0.728
Total       1.029      -  0.900  0.860  0.647  1.009  0.798

FB            out    $hr     $h    $xb    $3b   $roe    r/o
Offensive   1.019  1.160  0.773  0.996  1.038  0.350  1.002
Defensive   1.005  1.325  0.763  0.914  0.320  0.840  1.128
Total       1.012  1.242  0.768  0.955  0.679  0.595  1.065

LD            out    $hr     $h    $xb    $3b   $roe    r/o
Offensive   1.267  0.681  0.925  0.912  0.728  0.817  0.675
Defensive   1.124  0.698  0.966  0.699  0.528  0.854  0.808
Total       1.196  0.689  0.945  0.805  0.628  0.836  0.742

Dodger Stadium has benefited Dodger pitchers significantly in walks and strikeouts. It has been neutral for Dodgers hitters, so it’s easy to speculate that Dodger Stadium has a tough hitter’s backdrop that Dodger hitters get used to. In addition, it has, as expected, greatly increased the number of foul outs.

Dodger Stadium has been very tough on ground balls. Many more grounders are converted into outs and the number of extra base hits and triples is reduced. Thus, ground ball pitchers should benefit a great deal from Dodger Stadium, as some preliminary research has indicated. Don’t flip out about that ‘div/0’ in the triples column; there were only nine groundball triples in the data sample and none of them happened to have been allowed by Dodger pitchers on the road.

In the past, I’ve theorized that Dodger Stadium was a better park for fly ball pitchers because it reduces the number of doubles hit. That appears to be absolutely wrong, as while it does reduce the number of doubles on fly balls it does so by turning them into home runs; for the number of fly balls that become doubles, triples, and home runs combined Dodger Stadium has been neutral. By turning extra fly balls into home runs, Dodger Stadium increases the run value of a fly ball.

Most surprising, though, were the results for line drives. Can Dodger Stadium really be severely increasing the number of line drive outs? Not likely. Given that the park factor for line drives was 1.055, it’s fairly clear that the STATS scorers at Dodger Stadium have called more balls line drives than average. So whether the number of line drives that become home runs at Dodger Stadium is a result of the park’s dimensions or is due to the scorers at Chavez Ravine calling more home runs fly balls, I’m not sure. I can’t think of any method that tries to divide up the excess fly balls, so let’s take a look at the numbers if we combine line drives with fly balls:

air           $hr     $h    $xb    $3b   $roe    r/o   outs  xbh/air
Offensive   1.026  0.932  0.891  0.844  0.422  0.877  1.038    0.955
Defensive   1.142  0.980  0.712  0.419  0.862  0.979  1.002    0.847
Total       1.084  0.956  0.802  0.632  0.642  0.928  1.020    0.901

Once again, foul outs were removed from this calculation. As we would probably expect, Dodger Stadium hasn’t had a substantial effect on the number of line drives and fly balls that were caught for outs. It has turned a lot of balls into home runs, though, and Dodgers pitchers have been the main victims. At the same time, however, it’s reduced the overall number of extra base hits considerably, even when including home runs (that’s the xbh column at the end). The major difference between the rate of extra base hits on balls in play for Dodgers hitters and pitchers is not surprising, as that’s one of the more sizable areas of home field advantage in general.

Based on this data, then, it would appear that the extra base hits Dodger Stadium takes away are on balls with lower hang time, as the effects on line drives – though ostensibly including tweener balls that were classified line drives but may have been fly balls elsewhere – were much more severe than on fly balls. This makes some intuitive sense, as fly ball doubles would tend to result from balls that are simply hit far enough from fielders that by the time they’re fielded the batter’s already close to second base. Conversely, line drive doubles are probably much more dependent on funny bounces and wall angles, the kind of shenanigans Chavez Ravine’s very clean layout doesn’t tolerate.

Furthermore, while this evidence is far from conclusive on the matter, it lends some minuscule credence to the notion that Dodger Stadium keeps sharply hit balls from going over the fence but helps loftier fly balls turn into home runs. As such, this may be further evidence that groundball pitchers should benefit more from Dodger Stadium, as Robert Dudek’s excellent article on hang time in the THT Annual indicated that the fly balls that groundball pitchers give up tend to stay airborne for less time, as the fly balls they allow will often result from “mistake” pitches. Incidentally, recent Dodger signee Derek Lowe was one of the pitchers observed for Dudek’s study.

So how much of an impact on overall scoring does each component have? To give a rough idea, I ran the numbers for what Dodger Stadium would look like if I normalized each major component of the park’s impact on scoring. To do this, I set up all the offensive totals as a function of each $component, and to isolate the effect of one $component or group of $components I substituted the away value for that component into the home calculations. I divided the park’s effects into four categories: Foul Outs, Backdrop, Groundballs, Airballs. For foul outs, I normalized the $foul component. For Backdrop, I normalized $hbp, $bb, and $k. For Groundballs, I normalized groundball $h, $xb, $3b, and $roe. For Airballs, I normalized $hr, $h, $xb, $3b, and $roe for line drives and fly balls, excluding foul outs.

So here’s a rough idea of what Dodger Stadium’s Base Runs per out would be if it played as perfectly neutral for foul outs, hitter’s backdrop, groundballs, and airballs:

Now, here’s what it looks like the other way around – if everything is normalized except the component in question:

The overall impact of airballs is largest, though for the pitchers groundballs and hitter’s backdrop had a bigger impact. Clearly, the impact of foul outs has been, on the whole, fairly small. This might seem counter-intuitive at first, but when you consider that the loss of a foul out simply turns into another plate appearance and 68.8% of all non foul out PA’s at Dodger Stadium became outs anyway, it’s not hard to grasp that the overall impact is fairly marginal.

A story which may have evaded the notice of most non-Dodgers fans this season is the renovation at Dodgers Stadium. Seats have been removed from the outfield while new seats have been added where there used to be foul ground around the infield and behind home plate. It’s unclear, to me, that this should have any substantial effect on Chavez Ravine’s scoring environment aside from a reduction in the number of foul balls caught for outs. Actually, it should marginally affect the frequency of runners advancing on passed balls and wild pitches, but that’s beyond the scope of this research.

I’ve seen speculation that this renovation will have almost no effect; I’ve also seen speculation that this will turn Dodger Stadium into a neutral park or even a hitter’s park. The data I presented above certainly suggests that the impact will not be too large. Using my normalization model, I found that Dodger Stadium would have a park factor of about .92 if there were no foul outs recorded whatsoever. That’s clearly a figure in need of salt, as I imagine having no foul ground would impact the pitcher-batter confrontation in other ways. What that really measures is what things would look like if every foul ball was dropped.

How many foul outs can we expect to see next season? Using the seating chart at Dodgers.com, it looks like Dodger Stadium will have about 40% as much foul ground next season. If we assume the number of foul outs is directly proportional to the amount of foul ground – an assumption I can’t back up, but one I feel comfortable making nonetheless – that reduces the park factor for foul outs from 1.274 to 0.510. Using the above normalization method, that would bring the park factor for base runs per out up from the observed .866 to .899. That’s certainly a significant change, but it’s also a change small enough that random sample size noise will make it largely invisible.

The poster boy of sorts for the Dodger Stadium renovation is Derek Lowe, whose four year, $36 million signing this offseason surprised most sabermetrically-inclined onlookers. I’d prefer not to engage in idol speculation on how Lowe will perform with the Dodgers. However, we can fairly easily take Lowe’s performance record in Boston and translate it to a 2005 Dodgers context.

To make the translations, I used Lowe’s batted ball data (GB, FB, LD) and his TBF, HBP, BB, and K data. I applied park adjustments for Dodger Stadium. I did not use his actual single, double, triple, home run, and reached on error data. For the park adjustments, I used the values from this study, regressed to reflect the sample size, and applied them to the league averages for results of batted balls. I also used the Dodgers’ total park factors rather than the defense-only park factors to account for homefield advantage differential (in other words, to reflect that part of the defensive park factor is poorer defense on the road). I also assumed Lowe would face equal numbers of batters at home and on the road. I added in adjustments for double plays and the Dodgers defense; I based the Dodgers defense on regressed UZR data, with a +20 value for the Dodgers infield and +8 for the outfield (with Drew in center and Bradley in right).

Keep in mind, this is just based on batted ball classification (grounder, fly, line drive), not the distance or speed of batted balls. Salty salt salt salt.

Here are Derek Lowe’s numbers, translated for how they’d look in 2005 with Los Angeles. Note that this is run average, not earned run average. The baseline for RAA and RA+ was the league average starting pitcher, for whom I made the same translation for playing in Dodger Stadium with the same defense (the resulting RA was 4.42):

        RA  RA+     IP  RAA
2002  3.26  135  213.1   27
2003  3.72  119  213.0   17
2004  3.91  113  200.1   11

If Lowe can put up 11 RAA over 200 innings he’s probably worth the $7.5 million he’s owed in 2005, considering the size of the Dodgers’ payroll and their chances at the playoffs. And if he does that, I’d bet some team would be willing to give up a little too much for the right to pay the majority of the $28.5 million remaining on his contract. Paul DePodesta inking Lowe may be a move that we look back upon as very, very shrewd.