The Ballparks Project - Research Phase

[Amaroq]

Any PureSim fans out there game for maybe doing an organized creation of a number of PureSim .prk files so that others could benefit? It would be great if we had a place where folks could just download and import parks and be ready to go!

Any volunteers?

I started off down that path, for a while. This thread is about capturing that knowledge for anybody who is interested; my thought is that its more work than one fan will be able to accomplish alone but it might be a way to touch on it. The purpose of this thread is purely to provide knowledge for anybody who might want to undertake 'The Ballparks Project'... not to say "I'll help" or anything like that.

I. Variability - One thing I noticed was that ball park dimensions changed frequently. For example, its not sufficient to say "Forbes Field" - what we would really want is "Forbes Field 1909", "Forbes Field 1921", "Forbes Field 1926", "Forbes Field 1930", "Forbes Field 1947", "Forbes Field 1949", "Forbes Field 1950", and "Forbes Field 1954". That's a lot more work - but some of the ballparks changed a lot with the same name.

PureSim request: It would be nice if we could output "PSPNcast configurations" as one sharable output file, and "park dimensions" as another - I can imagine that most people would be content with a "Forbes Field PSPNcast" file, and might not care to go through all of the incarnations of Forbes Field over time.

II. Research - The second thing I noticed was that it isn't always possible to translate what somebody meant by the ballpark definition they've noted into a PureSim stadium map. Where is "Left Center"? Does "Right field" mean 'straight over the right fielder' or 'straight down the foul line'? This is especially true in some of the oldest ballparks, which are very spottily documented.

The following are useful research links for anybody contemplating this project:
Ballparks.com - N.L.
Ballparks.com - A.L.
Baseballguru.com - MLB parks since 1900

III. Real life != PureSim. The next thing I noticed was that the historically correct ball park dimensions don't lead to historically correct numbers in-game.

I have several theories about this -
- the outfield walls really only turn home runs into outs and vice versa (should also affect doubles, triples, etc.)
- a 'short porch' translates into good power numbers for the players... and then if you give them a short porch in the ballpark definition, they 'overperform' since their power is aimed at performing in the 'average' ballpark.
- some of the constants around turning an out into a home run and vice versa may be too powerful, especially for very extreme numbers (very short, very long)

I've got a spreadsheet with some reasonable guesstimations in it (taking into account II and III)... but I've found that it really takes playing with individual ballparks to get them to 'feel' right in terms of 'generating historically believable numbers'.

So, before we go down a ''lets build this" path, I think we'd need to decide whether we were building the 'exact correct' numbers, or whether we we build parks that subsequently generate 'historically believable' numbers.

PureSim request: could we tighten up the difference? For example, is it possible to go from Lahman to 'Sabremetric' numbers which take into account park factors, and thus prevent us from re-adding them?

IV. GUI limitations. At the moment there are some limits in the GUI - minimum wall height 5', maximum wall height 50', minimum wall depth 250', and maximum wall depth 475'. All four of those are exceeded by somebody trying to recreate all of the vagaries of MLB ballpark history. I've run into walls of 60', walls of 0' (done in protest to MLB complaining his walls were too high), depth of 235', and depth of over 600'.

PureSim request: for this to work, we'd need the GUI restrictions lifted... and if we're going to go down that route, we might want 'extreme' parks toned down a bit.

V. Obscure park features. It isn't possible to do odd, obscure park features like the scoreboard at Forbes Field or the statue inlet in deep center of the Polo Grounds.

PureSim request: we'd need support for this.

VI. Foul territory.

PureSim request: different parks have different ammounts of foul territory both along each foul line and behind home plate; its probably more worth your development time to add this than to do some of the other things mentioned earlier in the thread.

VII. Put it in the game?.

PureSim request: If we do the work to generate 'historically correct' ballparks, is there any way we could get them into the game? Or perhaps build a system by which a specific team follows the real ballpark dimension changes of their real-world counterpart? Maybe it isn't possible for Shaun to release 'Forbes Field' with the game, but if we've done all the work to generate Forbes Field info for each of the years 1909-1970, maybe we could have a mechanism for mapping which 'park dimensions' files to use in which seasons?

[spartacus007]

I've started and stopped that project a few times, running into the same problems you described. Even as late as the early 1900s there are some parks with no information at all. For my 1901 file I could find dimensions for only half the parks and for the one in Baltimore I couldn't find either a park or a photo.

With the newer parks I can look at the PureSim outline and compare with photos to check that I'm not way off in which numbers to put in which boxes. Even with parks that fall within the maximum and minimum dimensions, figuring out what numbers to put in 'LF3' or 'CF1.' If you used Ballparks.com like I did, you know just what I'm talking about. My Polo Grounds parks looked like some strange cookie cutter.

I'm all up for setting up a forum to get feedback on specific parks and eras. Other than finding a way around the dimension restrictions, I don't think there needs to be any programming on Shaun's part. I figure that if people are willing to find and download park files, they would probably enjoy changing the parks for the changed teams each season, to see how baseball evolves.

[SpharV2]

Yep, I ran into pretty much all of these issues when I put the list of park dimensions together for Shaun way back when, and ended up making a bunch of compromises. And your point 1 was a biggie, sometimes it came down to just finding things that might give the stadium a little more character. Another thing I'd like to see added would be more wall points so you could really get the outfield wall shapes. You can see how the Polo Grounds ended up looking because of the limited number of "anchors" for the outfield walls.

I don't want to sound like a whining customer (I know I hate when people say things like this, especially when they have no real idea of how much work it would be, like me) but if we could get a nice full featured editor for the parks, including most of the things above, like foul ground, more outfield anchors (I'd like to see at least 15-21), expanded limits and such, Shaun would never have to touch the thing again. I think we've got enough people here to put together a really nice list of parks. I know I'd be willing to put some more time into it.

[KG Erwin]

Here's some info on Forbes Field:

Dimensions: Left field: 360 (1909), 356.5 (1921), 356 (1922), 360 (1926), 365 (1930), 335 (1947), 365 (1954); deepest corner, left of straightaway center, at the flagpole: 462 (1909), 457 (1930); center field 442 (1926), 435 (1930); right-center, right side of exit gate: 416 (1955); right-center: 375 (1942); bend at left end of screen: 375; right field: 376 (1909), 376.5 (1921), 376 (1922), 300 (1925); backstop: 110 (1909), 84 (1938), 80 (1947), 84 (1953), 75 (1959).

Fences: Left-field front fence: 8 (5 screen above 3 wood, 1947), 12 (9 screen on top of 3 wood, 1949), 14 (screen, 1950); left-field wall: 12 (1909), 12 (brick and ivy, 1946); left-field scoreboard: 25.42 (steel left and right sides), 27 (middle); wooden marine sergeant at parade rest to right of scoreboard: 32 (June 26, 1943, to end of season); side wall angling back to meet brick wall in left-center: 12 (wood, when front fence was up); cages around light tower just right of scoreboard and in power alleys: 16.5; center field: 12 (wood, 1909), 12 (brick and ivy, 1946); right-center: 9.5 (concrete, 1925); screen - left side at 375 Mark: 24 (14.5 wire above 9.5 concrete, 1932); screen - right side at flagpole: 27.67 (18.17 wire above 9.5 concrete, 1932).

This info comes from http://www.ballparks.com/baseball/index.htm

[PadresFan104]

Yikes... You guys are hardcore!!! I'm just worried about about the graphics, but you are taking it to quite another level..

What I don't understand is, why take the time to put so much effort into this? Take historical replayers for example. The players statistics for each year already take into account the real fields they played on. How is recreating the exact dimensions in the game going to affect the sim? Do you just want the data for informational purposes? Like an encyclopedia?

For those that want to see how Babe Ruth would do at Minute Maid Field, or how Barry Bonds would do at the Polo Grounds, I don't see how PureSim, or any other game for that matter can simulate this. The Lahman DB doesn't store the length of homeruns, or any other hits, so I just don't see where this is going.

Do you all really expect the game engine to take into account the height of walls, and a slight change in field dimensions, when calculating if a hit is a double or triple or homerun?

It would seem to me that any affect that the fields had on the players from any time period are already built into their ratings and the best way to normalize players from one era playing in another era's fields would be to normalize against the stats from that era, and not take into account field differences.

Amaroq, I've read your post again, and I see that you make some similiar points, and you also have some valid ideas. I guess I'm feeling that this is too much work for a game like PureSim. Anything related to perfecting every aspect of the historical replay really seems better suited for a game like DMB.

I, for one, will be happy if Shaun just gives us more graphical tools for tweaking the ballpark layouts. [;)]

[puresimmer]

Do you all really expect the game engine to take into account the height of walls, and a slight change in field dimensions, when calculating if a hit is a double or triple or homerun?

Believe it or not it does

Try creating a park with 50ft walls and bring the fences in a bit and pretty soon you'll see some play by play that will mention a ball was "off the wall".

That said, I'd just be happy if someone would pull together some .prk files and we could start collecting them.

[PadresFan104]

ORIGINAL: puresimmer

Do you all really expect the game engine to take into account the height of walls, and a slight change in field dimensions, when calculating if a hit is a double or triple or homerun?

Believe it or not it does

Try creating a park with 50ft walls and bring the fences in a bit and pretty soon you'll see some play by play that will mention a ball was "off the wall".

And in the context of modern baseball with modern real or fictional players, this makes sense. But the folks here seem more concerned about the dimensions of the older parks for historical replay's.

I still contend that their stats already take the affects of the parks they played in into account.

Not trying to wizz on anyone's cornflakes, just looking for an explanation of why getting this information right is statistically significant.

Al

[Amaroq]

Not trying to wizz on anyone's cornflakes, just looking for an explanation of why getting this information right is statistically significant.

My cornflakes! My cornflakes! **looks up from his cereal bowl in horror** [;)]

For a historical replay played in the 'average park', in which the players spend their entire career in the correct stadiums .. yes, you're correct.

However... what if Babe Ruth had stayed in Boston, and played his entire career as a slugger with the Green Monster. What would that have done to his home run totals, career batting average, etc?

Even if you have him on the 'right' team, shouldn't he hit more home runs in New York (left field: 280ft, 4' fence in 1923, 301ft, 8' fence in 1928) and Boston (left field: 320ft, 25' fence from 1912-1933) than on the road in, say, Comiskey Park (363', 12' fence) or League Park (385', 5' fence) in Cleveland?

The particular 'flavor' of PureSim that I'm enjoying the most at the moment involves starting with the players from 1901, a 16-team association, and a 'dispersement draft', so that you might find Christy Mathewson on the Giants in one career, with the Red Sox in the next, and with the Browns in a third career. Even if I started with the 'right' teams from 1901, by the time I've had the amateur drafts of 1902-1906 and a few years of trades and free agency, the players have moved around a lot.. and the development of young players and aging of older players has diverged from 'historically true'.

By that point 'what park they play in' should have a larger effect on numbers like their home run totals than their 'Lahman' stats.

So, my ideal would be to have the player's ratings done 'normalized' for park effects, and let the ballpark dimensions generate effects better or worse for those players.

[SpharV2]

Well, speaking for myself, aside from just plain enjoying doing it, I like to be able to have lots of parks with character. Nothin worse to me than having a league like MLB in the 70s-80s with a ton of cookie cutter stadiums like Three Rivers and the Vet. Yuck.

Since I hardly ever play with real players, what it does to their baseline stats doesn't matter at all to me.

[PadresFan104]

ORIGINAL: Amaroq
However... what if Babe Ruth had stayed in Boston, and played his entire career as a slugger with the Green Monster. What would that have done to his home run totals, career batting average, etc?

Even if you have him on the 'right' team, shouldn't he hit more home runs in New York (left field: 280ft, 4' fence in 1923, 301ft, 8' fence in 1928) and Boston (left field: 320ft, 25' fence from 1912-1933) than on the road in, say, Comiskey Park (363', 12' fence) or League Park (385', 5' fence) in Cleveland?

Yeah, I see your logic. But to have the sim use these dimensions in any useful way, wouldn't we need to know the dimensions of the parks he played in, in order to normalize his stats? What if an historical player hit 10 HR's all to left field in a park with 330' fences at that spot. How is PureSim going to know this, and adjust his performance in other stadiums accordingly? I feel like I am missing something obvious, but it still sounds very arbitrary to me.

Now granted, PureSim is a ratings based game, not stat based. So if we are willing to throw stats out the window, and normalize players to the era they played in, then cool. I'd actually like to see that so we can have better matchups with players from different time periods. For example, a player with 15 homeruns from the deadball period would be rated high in power, about the same as a player with 45-50 homeruns in the modern age. Then you could throw stadiums in to the mix, and have park factors affect the outcome of each play.

[BryanK]

To take this in a completely different direction...

Anakit put together a set of ballpark photos for another baseball sim. There are pictures for every major league park except RFK, a handful of throwbacks, and like 3 generic stadiums. I've been toying around with them, and they look pretty slick when using something akin to Padresfan's PSPN template.
Theset pictures could go a long way toward a comprehensive stadium project; I'd be happy to clean them up a little and resize them for Puresim. I'd need a little help, though
1. I really don't want to configure the ball flight locations. Anyone want to do that? Given the way the stadium photos are oriented, I think it could be done in such a way as to only be done once. The PSPNcast stuff would also have to be configured, and I'm currently thinking I like Padresfan's setup.
2. For stadiums not already programmed into the game, we would need dimensions. I can't find dimensons with as much detail as Puresim requires. Stadiums that come to mind are The Great American Ballpark and Citizen's Bank.

Anyone game?

Couple additional notes...
1. Since I don't use MLB teams, I've already photoshopped out MLB logos for all but 2 stadiums. IMO, This makes them more useable.
2. The photos are screen shots from EA's MLB game. They are really nice, but they do look computer generated, and there are banner adds for EA products in some of the shots. Any potential rammifications for that?

[Amaroq]

But to have the sim use these dimensions in any useful way, wouldn't we need to know the dimensions of the parks he played in, in order to normalize his stats?

Not exactly, but you're very close to the mark.

There's a whole field (Sabermetrics) dedicated to normalizing stats. Three of the 'factors' that they seek to normalize back out are the fact that your player didn't face his own team, the 'park effect factor', and the 'league average'.

For example, there are two pitchers with a 4.00 ERA in the same number of innings pitched, with the same number of strikeouts, hits, and walks allowed. One played his home games in Coors Field, while the other played his home games in Turner Field. Which would you rather have? Well, if you could look at historical data for Coors Field and discover that, due to the altitude, its a great hitter's park, while Turner Field with its large foul territory and deep power alleys is a great pitcher's park, you'd realize that the Coors Field pitcher was objectively a much better pitcher.

Now, let's consider two .300 hitters. Both have the same number of AB's, hits, home runs, etc. One did so while playing for the Cincinnati Reds this season, while the other did so playing for the Houston Astros. Who was better? As it turns out, the Astros' pitching staff, with Clemens, Oswalt, and Pettit, was phenomenal, compiling a 3.51 ERA and 1164 strike outs; the Red's staff compiled a 5.15 ERA and 955 strike outs. Since the Astros player achieved his numbers without facing the Big Three, and the Reds player achieved his numbers without the benefit of facing his own team's abysmal pitching staff, one could conclude that the Reds' player had the better season.

Finally, let's consider two players who hit 7 home runs. One did so in the National League in 1909. The other did so in the National League in 1999. They had the same number of hits, at bats, etc... who had more power? The 1909 player led the league, while the 1999 player was nowhere near the league-leading total (65, Mark McGwire). So, the 1909 player was significantly above the average, while the 1999 player was significantly below the average.

The Sabermetric Manifesto

Obviously, these two factors interact .. but the basic idea is, if you are trying to create 'ratings' rather than 'a statistically accurate replay', the ratings should be based on the 'normalized' contact and power numbers for the player rather than the 'actual statistics'; and at that point, being able to build parks with the correct 'park effect factor' becomes very important.

[pasternakski]

Once you understand them, ball flight destinations are easy and only take a couple of minutes per ballpark. Aesthetic judgments are something you need to make your mind up about these before you start, particularly with regard to the "deep" and "x deep" outfield destinations. I keep these close to the fence, with the "x deep" a little higher up.

The whole process becomes pretty formulaic very quickly.

[PadresFan104]

ORIGINAL: Amaroq

But to have the sim use these dimensions in any useful way, wouldn't we need to know the dimensions of the parks he played in, in order to normalize his stats?

Not exactly, but you're very close to the mark.

There's a whole field (Sabermetrics) dedicated to normalizing stats. Three of the 'factors' that they seek to normalize back out are the fact that your player didn't face his own team, the 'park effect factor', and the 'league average'.

For example, there are two pitchers with a 4.00 ERA in the same number of innings pitched, with the same number of strikeouts, hits, and walks allowed. One played his home games in Coors Field, while the other played his home games in Turner Field. Which would you rather have? Well, if you could look at historical data for Coors Field and discover that, due to the altitude, its a great hitter's park, while Turner Field with its large foul territory and deep power alleys is a great pitcher's park, you'd realize that the Coors Field pitcher was objectively a much better pitcher.

Now, let's consider two .300 hitters. Both have the same number of AB's, hits, home runs, etc. One did so while playing for the Cincinnati Reds this season, while the other did so playing for the Houston Astros. Who was better? As it turns out, the Astros' pitching staff, with Clemens, Oswalt, and Pettit, was phenomenal, compiling a 3.51 ERA and 1164 strike outs; the Red's staff compiled a 5.15 ERA and 955 strike outs. Since the Astros player achieved his numbers without facing the Big Three, and the Reds player achieved his numbers without the benefit of facing his own team's abysmal pitching staff, one could conclude that the Reds' player had the better season.

Finally, let's consider two players who hit 7 home runs. One did so in the National League in 1909. The other did so in the National League in 1999. They had the same number of hits, at bats, etc... who had more power? The 1909 player led the league, while the 1999 player was nowhere near the league-leading total (65, Mark McGwire). So, the 1909 player was significantly above the average, while the 1999 player was significantly below the average.

The Sabermetric Manifesto

Obviously, these two factors interact .. but the basic idea is, if you are trying to create 'ratings' rather than 'a statistically accurate replay', the ratings should be based on the 'normalized' contact and power numbers for the player rather than the 'actual statistics'; and at that point, being able to build parks with the correct 'park effect factor' becomes very important.

Sounds good to me... Thanks for the added info!

Al

[PadresFan104]

ORIGINAL: BryanK

To take this in a completely different direction...

Anakit put together a set of ballpark photos for another baseball sim. There are pictures for every major league park except RFK, a handful of throwbacks, and like 3 generic stadiums. I've been toying around with them, and they look pretty slick when using something akin to Padresfan's PSPN template.
Theset pictures could go a long way toward a comprehensive stadium project; I'd be happy to clean them up a little and resize them for Puresim. I'd need a little help, though
1. I really don't want to configure the ball flight locations. Anyone want to do that? Given the way the stadium photos are oriented, I think it could be done in such a way as to only be done once. The PSPNcast stuff would also have to be configured, and I'm currently thinking I like Padresfan's setup.
2. For stadiums not already programmed into the game, we would need dimensions. I can't find dimensons with as much detail as Puresim requires. Stadiums that come to mind are The Great American Ballpark and Citizen's Bank.

Anyone game?

Couple additional notes...
1. Since I don't use MLB teams, I've already photoshopped out MLB logos for all but 2 stadiums. IMO, This makes them more useable.
2. The photos are screen shots from EA's MLB game. They are really nice, but they do look computer generated, and there are banner adds for EA products in some of the shots. Any potential rammifications for that?

Working on it! Should have a collection of current stadiums (with fence heights and all that) in a few days. The problem is that it uses my template, so if someone doesn't like the new look, it may not interest them. I'm using mostly photos that I found for use with the "other" game...

Al

[Amaroq]

2. The photos are screen shots from EA's MLB game. They are really nice, but they do look computer generated, and there are banner adds for EA products in some of the shots. Any potential rammifications for that?

There are copyright-violation issues with using screen shots of another product. (Screenshots are copyrighted if the displayed program is copyrighted - see wikimedia.org and jisclegal.ac.uk.) There is certainly no way that Shaun could incorporate or link to your work without violating EA's copyright, and its likely that your efforts (modification of copyrighted material, use of one coroporation's copyrighted material to enhance a competitor's product) would constitute an actionable violation of copyright law.

[:(]

[jearly]

I have a number of .prk files I've compiled for a few of my leagues. I have gotten the jpg's mostly from other members offerings, and have done some editting on a few. the file is a zipped file approx 10.6 Megs. If anyone is able to host it please let me know and as Shaun stated earlier - maybe we can get a decent compilation of parks.