In last week's post, I looked at Dan Uggla's advanced defensive statistics. The general conclusion was that these metrics uniformly agree that Uggla is a below-average defensive second baseman. More specifically, I also determined that the advanced stats were too unreliable to prove that Uggla's former home park had been harming his defensive statistics.
The question of whether Dolphins Stadium has some sort of defensive park effect remains open, however. To attempt to answer this question, I devised a fairly simple fielding stat that I call Defensive Out Per Year (or DOPY). In short, DOPY looks at the rate at which a fielder converts balls in his assigned range into outs. This rate is then converted into a raw number of outs based on the average number of ground balls and fly balls hit to that position in one year. Zero is set as the average; a total above zero indicates that a fielder converted more plays into outs than average.*
* I give a more thorough explanation of my methods at the bottom of the post, if you're interested.
The advantage of making my own system is that I can compare not just the home-road differences of the Marlins' 2B, but also the home-road differences of their opponents. If both groups of 2B do worse at Dolphins Stadium, then we'd have some real evidence of a park effect. If not, the theory would remain on tenuous ground.
After the jump, I graph these two data sets to see if we can reasonably expect Uggla's defense to improve now that he has escaped from
Jeffrey Loria's house of horrors Dolphins Stadium.
Overall DOPY ratings
Because we are mostly concerned with park effects here, I looked at all 2B on a team rather than individual players. For the Marlins, this doesn't make much difference, since Uggla was their second baseman for the vast majority of all plays from 2006 to 2010.
Overall, Marlins' 2B averaged a DOPY of -15.2 over the five seasons of Uggla's tenure there. That means that, given an average number of fly balls and ground balls, Marlins 2B would convert around 15 fewer balls into outs each season. (Do not compare DOPY to advanced metrics like UZR; those metrics are given in terms of runs saved, not outs made. These things are related, obviously, but not directly comparable.)
What about the Marlins' opponents? Second basemen playing against the Marlins averaged -6.7 DOPY from 2006-2010. So the Marlins' 2B were around 8 or 9 outs per year worse than their counterparts. Like most of the other stats we've seen (traditional and advanced), this tells us that Uggla was likely below-average defensively, though not catastrophic. Here's the breakdown for both groups by season:
The fact that even the Marlins' opponents had a negative DOPY is consistent with the idea that Dolphins Stadium is depressing the numbers of infielders, though it is obviously not proof. The real question, of course, is how the DOPY ratings at Dolphin Stadium compare to those at other stadiums.
Was Uggla really a worse fielder at home?
Let's look first at the home/road splits of Marlins' second basemen. Across the entire five-year period, the Marlins' 2B averaged -10.5 DOPY at home and -4.7 DOPY on the road. This difference indicates that Uggla and company were worse at home, which is consistent with the theory. However, the difference (less than 6 outs per year) is not particularly large, so we have to be careful. Also, we should not that the Marlins' 2B were still worse than average on the road.
Here is how the splits break down by year. Remember, these individual year splits are very small samples. I give this graph mainly to show my work and to allow you to see where any anomalies may be.
So the home-road gap existed only from 2007-2009, and was particularly strong in 2007. However, the trend seemed to reverse itself last year, as the Marlins' 2B actually converted plays in their range into outs at a much higher rate at home than on the road in 2010.
One thing that is interesting to me in this graph is the relative volatility of the home data, when compared to the road data. The year-to-year changes in the home DOPY values were consistently large: between 12 and 25 outs. The road DOPY values, on the other hand, had differences ranging only from 7 to 9 outs. If you think about it, this makes sense: the road values are split among a large number of parks, so any major differences in a given park will generally be evened out by data from other parks. Any changes to Dolphins Stadium, on the other hand, would have a large effect, since all the home games are played there (minus the few "home" games the Marlins played in Puerto Rico).
The other half of the equation
So far, the data provides a moderate amount of evidence to support the theory. But what about the Marlins' opponents? Did they also play worse at Dolphins Stadium?
In a word: no. The second basemen for the Marlins' opponents averaged a -0.8 DOPY at Dolphins Stadium; however, they had a -5.9 DOPY at their home parks when playing the Marlins. This means the opposing 2B converted about 5 more plays per year into outs when playing at Dolphins Stadium. Obviously, this contradicts the theory.
I should note that, because the Marlins' lineups have generally been dominated by right-handed hitters for the past 5 seasons, they hit fewer balls to the second baseman's range than other teams did. The Marlins hit about 2700 balls in this range during my sample; by comparison, the average MLB team had more than 3100. This means the sample size for the Marlins' opponents' 2B is somewhat smaller than the other data sets, and thus is a bit more volatile. However, this is still a pretty large sample, and all DOPY ratings are adjusted to the average number of GB and FB, so it shouldn't make a huge difference in the overall numbers.
The year-to-year data, however, is a different story. So all the caveats I applied to the last graph go double for this one:
Two things: first, there doesn't appear to be much of a pattern here. Second, it's very possible that the 2007 season just had some freaky data. If you leave the 2007 data out of it, the Marlins' opponents' 2B averaged -2.4 DOPY at Dolphins Stadium and -3.2 DOPY elsewhere. That's basically no difference. However, even this adjusted version is still evidence against the "Dolphins Stadium park effects" theory.
Wrapping up the Dolphins Stadium discussion
No matter how you slice it, there is no evidence that Dolphins Stadium hurts the fielding of second basemen in general. In fact, when you consider all 2B (and not just the Marlins), you get these numbers:
All 2B at Dolphins Stadium: -11.3 DOPY
Same 2B at other stadiums: -10.6 DOPY
According to my measurements, anyway, there is no park effect at Dolphins Stadium. I'm not saying you should definitely trust my measurements--I trust them a bit more than other defensive stats, but that's only because I know what went into them. You should view all defensive statistics with skepticism. DOPY is not the end of the conversation, just the best data that I could come up with. This does not disprove the idea that Uggla's poor fielding was due to his home ballpark, but it does weaken that idea--at least to the extent that you buy into my data.
Here's an infographic that summarizes everything DOPY tells us about the Marlins' 2B and the effects of Dolphins Stadium from 2006-2010 (click to embiggen if you want):
To prove a park effect, we'd want to see both the green bars be lower than both the white bars. But that's not what we see. Rather, as far as DOPY is concerned, whatever made the Marlins' 2B a bit worse at home did not apply to opposing 2B. The most likely explanation is definitely luck, but we can't rule out the hard-to-prove idea that there was some aspect of the field in Miami that hurt Uggla's fielding much more so than it did other fielders.
In summary, my analysis of the data for the Marlins and their opponents shows that:
- The Marlins' 2B (mainly Uggla) were about 15 outs/year below average (-15.2 DOPY).
- Their opponents' second basemen were about 7 outs/year below average (-6.7 DOPY).
- The Marlins' 2B were about 6 outs per year worse than at home than on the road, HOWEVER
- Opposing 2B were actually about 5 outs per year better at Dolphins Stadium.
- Overall, my measurements show no evidence of a park effect at Dolphins Stadium.
Oh, but one more thing...
Even if Uggla's defensive struggles had nothing to do with Dolphins Stadium, that is only really half of the equation that Braves fans should be interested in. We should also look at Turner Field to see if it has had any sort of park effect on second basemen. Accordingly, I ran the DOPY numbers for the Braves 2B and their opponents over the past 5 seasons. The results were very interesting.
Overall, the Braves' 2B came in at +14.2 DOPY from 2006-2010. That's a very good total. However, their opponents' 2B also fared well overall: +7.5 DOPY. At this point, I'll just jump to the infographic because it really tells the rest of the story:
I'm not saying that Turner Field definitely has a positive park effect on second basemen, but if it did... THAT is what such a park effect would look like. Both the Braves' 2B and their opponents' 2B did much better at Turner Field than they did elsewhere. What's more, this effect was the same for both groups. The Braves' 2B were 13.6 outs/year better at the Ted, and opponents' 2B were 13.1 outs/year better.
Given the magnitude of this difference and the fact that it was the same for both the Braves and their opponents, I would say that there is a good chance that Turner Field does have a real defensive park effect, at least for second basemen. I don't know what would cause this effect, but the best explanation I can come up with is that our grounds crew has specifically tailored the field to make it easier for second basemen to field grounders. This is a plausible scenario because the Braves have had some extreme groundballers on the staff over the years; the grounds crew should try to make these pitchers' jobs a bit easier.
Whatever the explanation, if this park effect is real, we would expect Uggla's fielding numbers to improve now that he's on the Braves. Of course, it won't be because he has improved, but rather because he'll be in a better fielding environment. It's sort of like how you'd expect a below-average hitter to seem better when he plays for the Rockies, or a below-average pitcher to seem better when he plays for the Padres. Except, of course, that those pitching/hitting effects are well established, and defensive park effects are still pretty dubious.
We will have a much better idea about all of this after Uggla has accrued a few seasons' worth of data at the Ted. Until then, his defense should remain a well-worn subplot. All I'm really hoping for is to have his fielding not become so bad that the Braves have to move him off of 2B. At Uggla's age, even if Turner Field really does hide some of his defensive flaws, it is far from certain that he'll be able to play 2B for five more years.
What is the methodology for calculating DOPY?
As promised, here is a more thorough explanation of DOPY. I use 2B as an example in this explanation, but you can also compute DOPY for other positions. Hopefully, this is comprehensible. If not, feel free to ask questions about it.
- Take all the ground balls hit in a fielder's range, and subtract those that were fielded by other infielders.
- Determine how many of the grounders were converted into outs.
- Add the number of ground ball double plays started to the values from Step 2. This is the total number of outs made.
- Take the total from Step 3 and divide it by the total from Step 1. This is the Ground Ball Out Rate.
- Repeat steps 1-4 for non-grounders (fly balls and line drives). The result is the Fly Ball Out Rate.
- Multiply the Ground Ball Out Rate by 480 and the Fly Ball Out Rate by 120. The purpose of this is to eliminate any biases caused by playing behind extreme GB or FB pitchers. The average # of balls hit to a 2B in one year is a bit above 600, and 80% of these are GB (80% of 600 is 480).
- Add the two values from Step 6 together. This is the unadjusted Defensive Outs Per Year.
- Finally, subtract the league-average DOPY from the player's value to get the adjusted DOPY, with 0 being average.
Here's an example, using the overall numbers for Marlins 2B:
- From 2006-2010, there were 2672 ground balls hit in the 2B range that were not fielded by other infielders.
- Of these, 1795 were converted into outs.
- Marlins 2B started 171 GB double plays, so that makes 1795 + 171 = 1966 total outs.
- The GB out rate is thus 1966 / 2672 or about 0.7358.
- The Marlins 2B had 586 fly balls or line drives hit to them, of which 520 became outs and 17 became FB double plays. That makes 537 total outs and 537 / 586 = 0.9164 FB out rate.
- The standard number of GB outs made is 480 * 0.7358 = 353.2; the standard number of FB outs is 120 * 0.9164 = 110.0.
- Add these together and you get 110.0 + 353.2 = 463.3 for the unadjusted DOPY (slight rounding difference there).
- Finally, subtract the league average, which is 478.5, from the value in Step 7 to get the adjusted DOPY: 463.3 - 478.5 = -15.2 DOPY, or about 15 outs less than average, per year.
I got all of this information from Baseball-Reference's Play Index, specifically the Event Finder. For instance, the total number of outs (used in step 2) can be found by going to "Batting by Team," then searching for "Any Batter Out" for all Teams from 2006-2010. Then, narrow the search to "vs. FLA", "fielded by 2B" and hit to locations "4" and "34". This restricts the search to outs that were made by Marlins 2B, in the second baseman's zone. (DOPY does not considered out-of-zone plays, but I found that that made little difference.) You can then do a search to break these plays down into GBs and FB/LDs, and another search to see how many were double plays.
For the total number of balls hit in range (the figure in Step 1), you do a similar search, this time for all "At-Bats". Restrict the range in the same ways as above, but add "fielded by RF" and "fielded by LF" to get the plays that the 2B missed.
To do the Marlins' opponents, restrict the search to "by FLA" instead of "vs. FLA". To do a home/road split, simply restrict the search to "Home" or "Away" plays, and multiply the rates by 240 and 60 (instead of 480 and 120). This is done so that the Home and Away DOPY values add up to the overall DOPY value.
One further note: The 2006 Event Finder data for balls hit in the 2B range but NOT fielded was incomplete, so I added a standard adjustment for all data sets in this year. This seemed to have little effect on the overall results.