Where have the Triples Gone, Part II?
A triple inspection double header.
As we continue to examine the causes of the decline in triples, one possible explanation that keeps resurfacing is the change in ballparks over time. This is certainly a logical concept to look into, and fortunately, it’s one that we can attempt to control for.
To do that, this study will look at those ballparks that haven’t changed significantly over a certain period of time (which are hard to find, I might add). It also will require game-by-game data to determine how many triples were hit in these parks. It’s extremely important to acknowledge the amazing work of those who provide this data for free so we can do studies like this, and I’d like to do that here.
- Retrosheet provided the game logs
necessary to do this study. There is a wealth of incredible information to
be found at their website, even for those just browsing around. I highly recommend
a visit.
- Sean Lahman’s Database provided
all of the data used in part one of this study—from triple rates to handedness—and
has been an invaluable resource for a lot of other work as well.
- Ballparks.com had readily available
the information I needed on park dimensions and changes. They also have pictures,
history, and other facts about the parks—not just for baseball, but other
sports as well.
The triple data I had available ran from 1974-2001, so it was necessary to find stadiums whose dimensioned remained relatively unchanged over that time period. Restricting the data to years from 1975-2000, I was able to find six suitable parks: Fenway Park, Wrigley Field, Milwaukee County Stadium, Veterans Stadium, Three Rivers Stadium, and Cinergy Field.
The following are the changes that have occurred at those parks:
- Fenway Park: Padding was added to the outfield wall in 1976.
- Wrigley Field: The backstop has varied in height from 60’ to 62.42’ over
the course of the study, and night play began in 1988.
- Milwaukee County Stadium: The left and centerfield fences were raised from
8.33’ to 10’ in 1985.
- Veterans Stadium: No changes.
- Three Rivers Stadium: Between 1982 and 1983, the surface was changed from
Tartanturf to Astroturf.
- Cinergy Field: The fences were lowered from 12’ to 8’ in 1984.
None of the above parks have had dimensional changes during the time period in question, to my knowledge.
Here is the triple rate at those six parks only, from 1975-2000:
3B/AB 3B/(AB-HR-SO) 3B/(H-HR) 3B/H Years
.0078 .0094 .0319 .0290 1975-1979
.0076 .0092 .0313 .0286 1980-1984
.0067 .0083 .0282 .0255 1985-1989
.0064 .0079 .0268 .0244 1990-1994
.0060 .0078 .0254 .0226 1995-2000
Clearly, there has been a large decline even in a relatively unchanging setting. Here is the rate of the league as a whole:
3B/AB 3B/(AB-HR-SO) 3B/(H-HR) 3B/H Years
.0074 .0089 .0310 .0284 1975-1979
.0070 .0085 .0295 .0270 1980-1984
.0062 .0077 .0267 .0240 1985-1989
.0061 .0075 .0259 .0235 1990-1994
.0056 .0072 .0237 .0209 1995-2000
While the overall decline is very similar, it has been a little greater. Here is the percent change between the beginning and ending periods for the two datasets:
3B/AB 3B/(AB-HR-SO) 3B/(H-HR) 3B/H Parks
-23.1% -17.0% -20.4% -22.1% Six parks only
-24.3% -19.1% -23.5% -26.4% All parks
I can’t say with any certainty that the difference between the two numbers is the effect that changing ballparks have had, but I think it is clear that the majority of change over this time period has been from other factors. Two other causes that have been suggested are improved outfield defense and taking the extra base being less worth the risk.
While I think the latter has had an effect, I think the first has been much more important. The reason is that not only would I expect that outfield defense has improved, and that that would reduce the triple rate, but that it would also follow the pattern that it has followed—decrease at a diminishing rate.
Dan 'The Boy' Werr
Posted: August 05, 2002 at 06:00 AM |
8 comment(s)
Login to Bookmark
Related News:
Reader Comments and Retorts
Go to end of page
Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.
1. Charles Saeger Posted: August 05, 2002 at 12:38 AM (#605729)Could you check the outfield assist rate and run a correlation? How about the flyout rate (PO-SO-A)? The percentage of flyouts the outfielders take?
Well, I would call those 2, 3, and 4, probably in the same order.
I guess I don't quite understand... to me it seems like something I'd clearly expect: outfielders get better as time goes on. It happens pretty much in every sport. In exactly the pattern Stephen Jay Gould describes in Full House: "For almost every sport, the improvement in absolute records follows a definite pattern... Improvement does not follow a linear path of constant rate. Rather, times and records fall more rapidly early in the sequence and then slow markedly, sometimes reaching a plateau of no further advance (or of minutest measurable increments from old records)."
So I would intuitively expect triple rates to decline, and have the rate of decline slow over time. The pattern in reality bears this out. It's not a definitive answer, but it does tell me that we're probably going on the right track.
And we can corroborate it, as I have done above. I have no doubt that the three causes you list have had some effect in the decline over the past century. Perhaps a substantial effect. But I doubt strongly that any have had a substantial effect in those six parks from 1975-2000, yet the triple rate has continued to decline at a rate near that of the whole league.
Let's look at it another way: assume the rate up above for the neutral parks is the true decline rate and it has been moving consistently. There are basically 20 years between the first and last sample. So let's assume that triple rates have been declining at that rate: .0064 3B/H per 20 years.
Here's the overall rate chart I posted on the other thread:
And then let's move back from the 1980s at that rate. .0255 in the 1980s means .0319 in the 1960s. Which means .0383 in the 1940s. Which means .0447 in the 1920s. So it's not until back then that the real rate passes the extrapolated park-neutral rate. And that's assuming that the rate of defensive improvement has been constant, when in reality we know that the rate should slow down.
Here's the same thing in chart form with the 90's at the start point and the rate divided in two: .0032 3B/AB every 10 years:
That's extremely quick and unscientific, but I still find it very revealing. Remember, the extrapolated rate on the left includes only those factors present in the six unchanging parks 1975-2000. And that the rate should actually be higher back when. And yet they suggest that triple rates from the 1930s on should actually be going down faster than they have been. And that even since the teens it accounts for 73% of the decline! Again, that's assuming the decline moved at the same rate back then, when it should be faster, raising that percentage.
Once again, it's not definitive, but I find it pretty darn compelling.
--
Could you check the outfield assist rate and run a correlation? How about the flyout rate (PO-SO-A)? The percentage of flyouts the outfielders take?
Charles, I'm not quite sure what you're getting at here. Is it related to your post at the end of the last thread? I'm guessing that the last two should be reversely correlated with triples under my theory. Not sure about the first...
However, the outfield assist rate (outfield assists as a percentage of H-HR+BB+HBP+0.6*Errors times PO-A-SO) is probably more telling. It's down, of course, but I want to know about an r coefficient.
So I would intuitively expect triple rates to decline, and have the rate of decline slow over time. The pattern in reality bears this out. It's not a definitive answer, but it does tell me that we're probably going on the right track.
Dan, I'm not really sure how you are extrapolating from Gould's hypothesis about improvement and declining rate of variation to an absolute decline rate. It sure seems as though you are misapplying what Gould said here. (Not that Gould managed to take anything into account about second-order evolution in the game, such as the rise in secondary average as a compensating factor for declining absolutes and declining variation in batting average.)
It's clear that the major portion of the decline in triples (a little more than 80% of it, in fact) occurred from the 20s to the 60s, and this is the area where additional information is needed in order to identify and isolate the factors involved in that decline.
Trying to use data from the more recent past, with a much smaller decline rate, in an age where parks were far more uniform in general than what was the case in the pre-expansion years, just doesn?t strike me as dealing with the fundamental issue here.
Trying to isolate the three factors from the 20s-60s would be the real piece of work to be done here. Rather than focusing only parks that haven?t changed, the extent of ballpark variation (both in terms of ballpark dimensions and park factors for triples) needs to be quantified.
You also mislabel the stat you?re using in your backwards extrapolation. You are using 3B/H, not 3B/AB.
One thing that might really help here would be to create a table/chart that uses running five-year windows for 3B/AB. By doing that, you might find some non-linear patches in the data and/or some steeper descent curves than what a decade-by-decade summary tends to create.
Nothing to add, just wanted to thank you for the great work. These articles were well written and thought provoking.
On his way to declining variation, Gould went through basic improvement, which is what I'm talking about here. He described the pattern it usually takes. If I'm correct that triple rates declining are a function of improvement, the pattern should hold (but of course be reversed). That's what I see. Once again, I'm not going into Gould's work on declining variation; I'm just using one of the same principles he used on his way to his conclusion.
Trying to use data from the more recent past, with a much smaller decline rate, in an age where parks were far more uniform in general than what was the case in the pre-expansion years, just doesn?t strike me as dealing with the fundamental issue here.
Well, I agree that it could be better. However, I'm using that to show that triples are declining from causes besides parks (not that parks aren't causing decline, but that something else also is). I see no reason to assume it's a new phenomenon if it is outfield defense improvement, and I don't know what else it could be, and it seems to me that defense would be improving and reducing the triple rate. It also seems to me that the rate of redcution should be declining, which is shown by the data. And even if it didn't, the rate it's going on now still shows a fair amount of the decline above.
In other words, if I found some massive park effect, I'd be pretty confused, because I'd have to ask why the defense isn't improving.
You also mislabel the stat you?re using in your backwards extrapolation. You are using 3B/H, not 3B/AB.
Good call. That was just a screwup on my part (I got it right above).
One thing that might really help here would be to create a table/chart that uses running five-year windows for 3B/AB. By doing that, you might find some non-linear patches in the data and/or some steeper descent curves than what a decade-by-decade summary tends to create.
Are you talking about like a moving average? I have no doubt that the data will be a little bumpy; there have been many other influences... I don't dispute that. Perhaps I'll do that though, it's probably a good idea. First I'll have to reorganize the data I have.
By the way, Don, I'm glad to see you posting your critique here. I appreciate your input and hate to miss it just by not stopping by your site for a week.
Charles,
I'll try to get to that, too. This has certainly gone far more in depth than I originally expected when I was looking at the rates and thought, "Hey, that's kind of interesting."
Brian,
Thanks... I can't tell you how good it is to hear that.
Thanks to everyone for all the good feedback.
You must be Registered and Logged In to post comments.
<< Back to main