Improvements to league evolution feature

[Le Grande Orange]

Originally Posted by Nunyer

+1 on the ability to regionally lock the automated expansion. When my All-Michigan league suddenly all had to deal with scheduling road trips to Fresno... well, yea...

There may be a perhaps relatively simple way to prevent this. If the game tallied up the typical population of the current members of the league, and measured the farthest geographic distribution of the member cities (which it could do since the latitude and longitude for each city is a part of its data), then it ought to be possible to tell the game to only select cities that are of comparable size to current members and that are either inside the current geographic boundaries of the league, or just outside them.

Originally Posted by Nunyer

Also would be nice to simulate an owners vote for changes as well.

I've been pitching the idea of 'interactive winter meetings' for two years now I think.

Basically, instead of league changes simply happening, as is the case now, there would instead be news stories generated detailing the list of proposed changes. This informs the user about what's upcoming. Then at the winter meeting each proposal comes up for a vote, and the user casts their vote on behalf of their team while the AI teams cast their votes. Each proposal is then either approved or rejected accordingly. There would then be news stories afterwards detailing the outcomes. (If you look at old issues of TSN or newspapers, you'll find exactly this sort of coverage of MLB's winter meetings. I've garnered some of the rule changes that have happened in earlier times from this coverage.)

Some changes currently are really just arbitrary almost to the point of silliness—like the DL being changed from 15 days to 13 days for example. I would argue any such sort of rule changes should be based more in reality, reflecting the kinds and degree of change that has happened in the real world. (The DL lengths in real life have been 7, 10, 15, 21, 30, and 60 days. OOTP should stick to only changing it to these values from the existing one, and IMO would only move one step in one direction or another. E.g. if at 15 days the rule change proposal would be to either shorten it to 10 days or lengthen it to 21 days.)

Ideally, changes would be based in some way on what's going on within the league itself, to make it feel more that the proposed change is in response to something that's happening rather than just a dice roll by the CPU.

[David Watts]

Originally Posted by Le Grande Orange

As the parent of this particular feature (which was originally inspired by something in Front Office Football), let me explain the rationale.

This feature was intended for those who want to have a universe which undergoes changes without their prompting. That is, instead of being the god of your universe and having to decide on everything that happens, you can instead sit back and go along for the ride—the story unfolds without you having to decide on every single plot point. It makes the baseball universe feel more 'alive' in that without such unprompted evolution the league would remain forever static and unchanging, which is not particularly realistic.

The league evolution which is in OOTP at the moment is just barely scratching the surface of what this feature could ultimately do. I have a clear vision of where it can go and what it could include... whether it'll ever get there is another matter entirely.

Please please please don't give up on your quest to improve and expand this feature.

[Le Grande Orange]

Originally Posted by David Watts

Please please please don't give up on your quest to improve and expand this feature.

Between the research, design, writing, and general contemplation of all the many ideas I have for this game, it could literally be a full time job, that's how much is involved. Unfortunately I have as yet failed to convince Markus to actually hire me as a full time feature concept creator/designer/researcher for the company.

[Questdog]

Originally Posted by Le Grande Orange

There may be a perhaps relatively simple way to prevent this. If the game tallied up the typical population of the current members of the league, and measured the farthest geographic distribution of the member cities (which it could do since the latitude and longitude for each city is a part of its data), then it ought to be possible to tell the game to only select cities that are of comparable size to current members and that are either inside the current geographic boundaries of the league, or just outside them.

To do this, though, you need a population by timeline, because the cities that are big and prosperous now are not the same ones that were big and prosperous 100 years ago......kind of silly when a team moves to Miami in 1880 (they had 257 people living in the area - not counting any Seminoles)......I have compiled just such a list for the USA and a handful of cities in Canada....

[David Watts]

Originally Posted by Le Grande Orange

Between the research, design, writing, and general contemplation of all the many ideas I have for this game, it could literally be a full time job, that's how much is involved. Unfortunately I have as yet failed to convince Markus to actually hire me as a full time feature concept creator/designer/researcher for the company.

Well I will keep my fingers crossed that he eventually comes to his sense

[Gambo]

Just make it able to be used in a "historical" league. So silly this isn't an option in league setup.

[Le Grande Orange]

Originally Posted by Questdog

To do this, though, you need a population by timeline, because the cities that are big and prosperous now are not the same ones that were big and prosperous 100 years ago...

Ideally, yes, but in the short term simply using the population figures currently in the database should be all right. (As a side note, minor leagues used to be classified based on the aggregate population of the member cities. Adapting such a system for use in OOTP to handle the relocation or expansion candidate cities would seem a good approach, especially when combined with geographical constraints and the level of the league.)

Originally Posted by Questdog

..I have compiled just such a list for the USA and a handful of cities in Canada....

I've combed through the U.S. Census Bureau site too. Some time ago I found the file listing the 100 largest U.S. cities in each census up to 1990. I've also found the population figures for each county in the U.S. for every census, and lots of other fun stuff too. Someday I'll take all that and work it into a market ranking system for U.S. cities from 1870 to the present...

[Questdog]

Originally Posted by Le Grande Orange

Ideally, yes, but in the short term simply using the population figures currently in the database should be all right. (As a side note, minor leagues used to be classified based on the aggregate population of the member cities. Adapting such a system for use in OOTP to handle the relocation or expansion candidate cities would seem a good approach, especially when combined with geographical constraints and the level of the league.)

I've combed through the U.S. Census Bureau site too. Some time ago I found the file listing the 100 largest U.S. cities in each census up to 1990. I've also found the population figures for each county in the U.S. for every census, and lots of other fun stuff too. Someday I'll take all that and work it into a market ranking system for U.S. cities from 1870 to the present...

I've already done it...........what I did is take the metropolitan (and similar) areas and figured their populations in all years based on the current makeup.....though in the datatbase I have retained the individual county populations in case I think of a better way to handle it.....it was a big job that had to be done by hand.....also, the Canadian info is harder to come by....I managed to get a lot of what was needed by making phone calls, but the info isn't as complete as the US data.....

[VanillaGorilla]

Add contraction to league evolution. Ty.

[Le Grande Orange]

Originally Posted by Questdog

I've already done it...........what I did is take the metropolitan (and similar) areas and figured their populations in all years based on the current makeup...

The issue there is that the metropolitan statistical area definitions have changed quite a bit over the years. For example, the Atlanta SMSA was first defined in 1950 and had only three constituent counties. In the 1960 and 1971 definitions it contained five counties. In 1981 it was up to fifteen counties. In 1990, eighteen counties; in 1999, twenty. The 2009 definitions gave the Atlanta SMSA twenty-eight counties. So which definitions best represent a metro area? If one only uses the current definitions that's probably overstating the effective population of the area in earlier years.

It's a tricky question, and one I've wrestled with for a long time. I'm leaning to a simpler system based purely on distance from the city, but that entails a lot of running a lot of calculations in Excel to get each county assigned to a metropolitan city area. I really don't have the stomach for that. (Some smart person no doubt could write a bit of software that could easily accomplish this task; alas, I am not that smart person.)

Originally Posted by Questdog

... though in the datatbase I have retained the individual county populations in case I think of a better way to handle it...

I agree. I've settled on a county-based approach for any market ranking/rating system. It has a number of advantages.

Originally Posted by Questdog

...also, the Canadian info is harder to come by...

Yeah, the U.S. Census Bureau web site is really outstanding. There is a vast amount of information easily accessible, and for free too. In Canada, not as much. To add to the matter there aren't really equivalents in Canada to American counties; the best proxy are either census divisions or subdivisions, depending on the province.

[Questdog]

Well, Atlanta is really about the only city that has that problem and I'm not sure it's a problem anyway.....certainly Atlanta was the center of activity in those 28 counties in 1880, if not where most of the people worked.......distance doesn't work so well in highly congested areas where there are several metro areas abutting one another.....no matter how you do you it, the solution will have to be a compromise of some sort and so far, I think using current definitions is the best answer.....most of the counties that weren't in an area in the past were only not in it out of some technical shortcoming, as the % of people working in the nearby city fell short, but since there were fewer commuters back then, it doesn't make sense to exclude those areas from being able to help support a ball club......also, the definitions of what are metro areas have changed, so it is not consistent to use whatever the definition at the time was.....but, most importantly, using my method is a huge step forward from just using city populations.....Boston, for one, is unfairly penalized for not annexing all the surrounding communities.....

[The Wolf]

Originally Posted by Le Grande Orange

Between the research, design, writing, and general contemplation of all the many ideas I have for this game, it could literally be a full time job, that's how much is involved. Unfortunately I have as yet failed to convince Markus to actually hire me as a full time feature concept creator/designer/researcher for the company.

He should.

[Le Grande Orange]

Originally Posted by Questdog

Well, Atlanta is really about the only city that has that problem ...

Not really. SMSA growth occurs throughout many of them. Note too that there were that SMSAs were created in 1950; and micropolitan statistical areas only started in 2003 I think it was. If you look back, for example, Dallas and Fort Worth were at one time defined as two separate SMSAs, and were later combined into one larger one.

Originally Posted by Questdog

...distance doesn't work so well in highly congested areas where there are several metro areas abutting one another...

True, but on the other hand, there is research that population within a given distance of the ballpark is a key component in potential attendance. This Baseball Prospectus study on the minors, for example, found that the amount of population within 10 miles of the ballpark was the most important factor for attendance. There was another study I found on JSTOR which found that attendance at English football games was also determined primarily by population within 10 miles of the stadium. It's on this basis I think a distance-based approach is worth considering. It also has the advantage of being a consistent measure across all time frames, unlike SMSA boundaries.

However, as you note, there is the question of closely-located cities. I've been using a 30-mile radius around the city as the distance for a major league 'market'. This distance was chosen for a couple of reasons, one of which is that the counties whose centers lie within 30 miles of the major league ballpark also tend to correlate highly with the counties that are defined as being that MLB club's exclusive operating territory. (For simplicity's sake I would advocate making territory and market size synonymous.)

Originally Posted by Questdog

...no matter how you do you it, the solution will have to be a compromise of some sort and so far, I think using current definitions is the best answer...

At one point I was going to pick a SMSA defintion that was somewhere in-between the first ones of 1950 and the most recent ones as a compromise. I've subsequently decided I prefer the distance model.

Originally Posted by Questdog

... using my method is a huge step forward from just using city populations... Boston, for one, is unfairly penalized for not annexing all the surrounding communities...

Pittsburgh is another city with a large difference between its city population and metro area population.

However, up to about 1950 if you ranked cities by the city population (i.e. the population within the legally incorporated boundaries of the city) and then by larger urban area population, the two lists were nearly identical. It's only after 1950 and the rise of the suburbs that the two lists begin to often diverge. (According to the Census Bureau, at any rate.)

[nyy26wc]

Originally Posted by David Watts

How about instead of More/less offense or More/less pitching if we had the option to choose a decade or more and have OOTP randomly cycle the modifiers through that time period. This way we could have changes, but wouldn't have to be afraid of the league turning into roid ball.

While we're on the subject of "more/less offense" and "more/less pitching", does anyone know the difference between the 2 of those?

I don't know if I should have both of them checked for when I create a hybrid fictional/historical player league or whether I should choose 1.

[Questdog]

Originally Posted by Le Grande Orange

True, but on the other hand, there is research that population within a given distance of the ballpark is a key component in potential attendance. This Baseball Prospectus study on the minors, for example, found that the amount of population within 10 miles of the ballpark was the most important factor for attendance. There was another study I found on JSTOR which found that attendance at English football games was also determined primarily by population within 10 miles of the stadium. It's on this basis I think a distance-based approach is worth considering. It also has the advantage of being a consistent measure across all time frames, unlike SMSA boundaries.

Of course, but finding the population with a 30 mile radius or 10-mile radius is impossible......your circle will have parts of counties that there is no way to get the data

Using current definitions across time is also consistent and I think a reasonable way to do it that is actually doable.....the lists using my method and just city populations before 1950 are not interchangeable....and using the different areas for different years makes no sense either, since the metro areas are mostly about finding where the people who work in an area live and including their homes and workplaces in the same statistical area so that federal funds are distributed in a manner that actually coincides with the amount of people affected......does the fact that people 20 miles outside of a city work in the city nowadays and worked in their own counties (on farms usually) 100 years ago have anything to do with whether that population can help support a ballclub?.....I don't think so.....so using current definitions of metro areas to cover all years gives a geographical stable area to consider (something even just city populations fails to do)......

[Questdog]

Originally Posted by nyy26wc

While we're on the subject of "more/less offense" and "more/less pitching", does anyone know the difference between the 2 of those?

I don't know if I should have both of them checked for when I create a hybrid fictional/historical player league or whether I should choose 1.

probably means changes to PCM's affecting either hitters or pitchers, but I don't know for sure.....never play with the feature turned on and don't really know how it works....

[Le Grande Orange]

Originally Posted by Questdog

Of course, but finding the population with a 30 mile radius or 10-mile radius is impossible......your circle will have parts of counties that there is no way to get the data

Which is why I said the counties whose centers lie within 30 miles (or whatever chosen distance). The county is either included as whole or it is excluded. It's an approximation to be sure, but from the tests I did the total population of the counties within 30 miles of a ML park was on average within 4% of the total population with 30 miles of the park without regard to counties (i.e. using census tracts or census block groups). Of course, some were more overstated or understated, but on average it was quite good. Certainly at a 10 or 15 mile radius there was a considerable difference between county and actual population.

(The Census Bureau conveniently published a list of the geographic centers of population for every county in the U.S. for both the 2000 and 2010 censuses.)

Originally Posted by Questdog

...does the fact that people 20 miles outside of a city work in the city nowadays and worked in their own counties (on farms usually) 100 years ago have anything to do with whether that population can help support a ballclub?.....I don't think so....

An argument in favour of the distance model, it seems to me.

Originally Posted by Questdog

...so using current definitions of metro areas to cover all years gives a geographical stable area to consider (something even just city populations fails to do)..

So does a distance model.

Something else to note is that I wasn't going to just factor in population, but also per capita income. Every study I've come across says the ability of a ball club to generate revenue is tied both to the population and income level of its surrounding city/urban area.

The BEA has population and per capita income estimates for every county in the U.S. from 1969-2012. For years before that, the Census Bureau has per capita income figures for 1959. I was going to just do a straight-line interpolation for the years between 1959 and 1969. For years prior to 1959, there really isn't much income data to work with in the census. I was going to use the BEA's estimates of state per capita income, which go back to 1929, as a proxy and adjust the county incomes in proportion to the changes the state underwent. A rough approximation, certainly, but the best that can be done with the available data. For years prior to 1929, I came a study which had revised estimates for state per capita incomes from 1880-1920. That would be used in a manner identical to that for the BEA's state income estimates.

[Questdog]

You may have problems in Western states where counties can be quite large and will make your theoretical areas that are supposed to be similar sizes, not very similar at all......by and large, I think you'll find as I did, once you actually get in to trying to accomplish it that using my method is a whole lot easier and no less accurate a measure......and will end up using most of the same counties in the end....

I think per capita income is important nowadays but not so much back when tickets were very cheap......your average Brooklyn fan of 1940 would not feel very comfortable at a Dodger game today.....the fans have made the same sort of economic transformation as the players, though on a smaller scale.....ticket purchases pre-free agency had more to do with desire to attend a game than affordability, I think......

[Le Grande Orange]

Originally Posted by Questdog

You may have problems in Western states where counties can be quite large and will make your theoretical areas that are supposed to be similar sizes, not very similar at all...

Never said it was perfect. Unless the system is only ranking things based on current-day data, there'll always be trade offs in making it historically authentic.

Originally Posted by Questdog

... and will end up using most of the same counties in the end...

I disagree. SMSAs tend to range out quite a ways in some cases.

Originally Posted by Questdog

...ticket purchases pre-free agency had more to do with desire to attend a game than affordability, I think......

And yet per game average attendance is much higher today than back then. Indeed, attendance has grown at a faster rate than the rise in the general population. (I've never tried graphing MLB average attendance against per capita income... I wonder if it'd correlate better.)

[Questdog]

Originally Posted by Le Grande Orange

I disagree. SMSAs tend to range out quite a ways in some cases.

SMSA is an outdated term.........the areas I've used are the smallest areas defined by the Census Bureau, not the CSA's....they are either metro- or micro- areas and are generally about the size of your 30-mile radius for large cities (major league towns)

Originally Posted by Le Grande Orange

And yet per game average attendance is much higher today than back then. Indeed, attendance has grown at a faster rate than the rise in the general population. (I've never tried graphing MLB average attendance against per capita income... I wonder if it'd correlate better.)

One other point: this doesn't need to be precise. My method gives a ranking of the Cities that seems very believable (much more so than using city population) and putting more work into it will certainly show diminishing returns....

And the biggest influence on overall attendance is the level of offense. When the balls are flying, the people are buying.....