2019 FBS imperialism map

For those who don't know or don't remember, FBS imperialism assigns land to each school. Basically, it's all of the counties that are closest to said school. Then, as teams win and lose, land shifts ownership. A guy on reddit handles it, but for some reason this year it hasn't been tweeted out much so I keep forgetting about it.

I found the map as of last week:

Obviously, VT immediately surrendered their original land to Boston College. Then, it passed through Kansas, WVU, Texas, Oklahoma, Kansas State, back to Texas, and currently at Iowa State. Kansas has a chance to take it again next weekend.

As for any land VT might have taken at any point this year? That is limited to ODU's in week 2, which we later surrendered to Duke. And as fate has it, we don't have any opportunities for land in the regular season.

DISCLAIMER: Forum topics may not have been written or edited by The Key Play staff.

Comments

This was fun the first half of the first year they did it. Then it was a little infuriating if you took it too serious (yeah, I remember taking it too serious)

What would be cool if they came up with a weighting system for losing land (more land lost for a blowout, more land lost for a home game, etc) rather than binary. I'm thinking something like the culture dependent city radius in Civ.

Warning: this post occasionally contains strong language (which may be unsuitable for children), unusual humor (which may be unsuitable for adults), and advanced mathematics (which may be unsuitable for liberal-arts majors)..

OK, so here's how I think it should be done.

Every team starts with their home county only. So, a lot of unclaimed, neutral land at the beginning.

Every week, you are assigned a score based on how your game went.
BYE: 2 points
win at home: +1
win on the road/neutral: +2
win, but lose against the spread: +0
beat the spread by 0-7 points: +1
beat the spread by 7-14 points: +2
beat the spread by more than 14 points: +3
lose on the road/neutral: 0
lose at home: -1
lose against the spread by more than 14 points: -1

You calculate the score and assign it to each of your counties. Compare the county score to any counties it touches (neutral gets a score of 0).
The county with the higher score takes over the other and the new county score is 1 less than the team score for the week.
Ties result in no change.
Repeat until all counties are ties. If a county could be taken over by multiple teams, the team who did better against the spread gets it. If there is still a tie, then RNG for winner.
Neutral cannot take over your home county.
At any point, if your team score is greater than the team score of the team holding your home county, you can retake the home county (and any surrounding counties that meet the criteria).

So, first week, you could gain up to 4 counties in every direction for a blowout road win (assuming all the counties are neutral). Later in the season, you could get up to 2 more counties if your are expanding into an area controlled by a team that was blown out at home.

The fun thing about this method is that, given our proximity to LOLUVa, we would have the opportunity to take them over early and own their county for the rest of the season, much like real life.

Warning: this post occasionally contains strong language (which may be unsuitable for children), unusual humor (which may be unsuitable for adults), and advanced mathematics (which may be unsuitable for liberal-arts majors)..

I foresee us having quite a varied empire, with parts of the northern Rockies, southern Appalachia, parts of Texas, and central North Carolina

Never Forget #1 Overall Seed UVA 54, #64 UMBC 74

How does Alabama not have any territory.

Overheard as Duke assistant coaches took elevator down from press box: “Guys, they stopped the run with a three-man front.” - David Teel Tweet 2018

They just lost two weeks ago.

Actually they lost literally when this map came out

If you lose the person that beat you gets all of your territory so Bama lost all the territory they had to LFSU and also didn't get and back this week because Miss St had no land to take

(add if applicable) /s

If you lose the person that beat you gets all of your territory so Bama lost all the territory they had to FSU and also didn't get and back this week because Miss St had no land to take

LSU

Now finish up them taters; I'm gonna go fondle my sweaters.

winning this map correlates to being the best CFB team as much as winning ToP correlates to winning your game...

Warning: this post occasionally contains strong language (which may be unsuitable for children), unusual humor (which may be unsuitable for adults), and advanced mathematics (which may be unsuitable for liberal-arts majors)..

There it is, statistical proof!

Dat R-squared doe...

Yes, but time of possessions gets us no territories, and we NEED all the territories.

@hokie_rd

that's kinda a weird scaling. Winning the TOP battle and W/L is more binary than a continuum. I don't think anyone has ever said: if we win the TOP by a greater amount, we'll win the game by a greater amount.

🦃 🦃 🦃

yeah, that's why I added the red and green boxes. It's about 55% chance that if you won ToP at all, you'd win the game. So, look at ToP and then flip a coin and call head's as the team that one ToP and you'll be right the same amount.

Warning: this post occasionally contains strong language (which may be unsuitable for children), unusual humor (which may be unsuitable for adults), and advanced mathematics (which may be unsuitable for liberal-arts majors)..

well, that's more than a coin flip, but not much more. Seems like it would be significant though, based on the number of games examined.

I look at this way. TOP can be factor in winning a game. In most games, it's more likely indicative of how the game went and not how the game was won.

🦃 🦃 🦃

I actually expected to see that blowouts correlated with large ToP differentials since teams that are winning tend to run the ball to burn off clock, but that wasn't even apparent. Admittedly this is just one season's worth of data (2005 I think?) because that's all that I could grab for free when I made it, but you still think there would be some top end correlation.

Warning: this post occasionally contains strong language (which may be unsuitable for children), unusual humor (which may be unsuitable for adults), and advanced mathematics (which may be unsuitable for liberal-arts majors)..

The model you built doesn't make sense if that's what you wanted to say. Why include a graphic with a regression line and an R squared on a continuum if you wanted to say that the percentage is the same? There are actual tests that could show if winning the ToP increases your chance of victory. Did you just want to show that they are about 25% correlated?

...That said I agree with the original point. I'd guess having more territory is correlated to more wins, but not as strongly as you would think. The team with the most territory after the first year was Pitt...because they beat previously undefeated Miami in their season finale, but weren't good enough to make a bowl game.

yeah, the graphic kind of evolved. The original was just trying to examine a relationship between ToP differential and score differential since the ongoing argument was whether ToP or points per play was a more valuable statistic. So the initial graph was side by side with this one:

Then I was lazy so I didn't rework it, just photoshop-ed some colors on it.

Warning: this post occasionally contains strong language (which may be unsuitable for children), unusual humor (which may be unsuitable for adults), and advanced mathematics (which may be unsuitable for liberal-arts majors)..

I've seen this graph and the underlying analysis - to be clear, this is a descriptive analysis, which I don't find valuable. If the goal is to create a game winning strategy, a prescriptive analysis is necessary, though one could make an argument that a predictive analysis could also be valuable.

When this discussion comes up, I always default back to my IE/OR roots and ask, "Given the data, how do you devise a strategy for generating a win". When asking this question, the points-per-play analysis does not provide a prescriptive strategy.

I'd accept the argument that TOP doesn't provide a prescriptive strategy either, though I'd also argue that it is possible to devise a strategy that includes a time of possession as a primary component, but you can't devise a strategy that seeks to maximize a points per play differential as primary component.

This is why I see both of the graphs above and still do not believe in points per play. I'm not interested in a descriptive correlation, I'm interested in making a prescriptive strategy.

As I recall, the original point was not which statistic could you make a game plan around, it was which statistic describes how well the team is doing. Joel was pointing out that ToP had little correlation with winning, but ppp was a useful statistic in that it lined up with winning. Yes, that's descriptive, not prescriptive. It doesn't matter if you're not interested in that.

Warning: this post occasionally contains strong language (which may be unsuitable for children), unusual humor (which may be unsuitable for adults), and advanced mathematics (which may be unsuitable for liberal-arts majors)..

IIRC, the description said that ppp differential was the best predictor of winning - my distaste for the concept is due to its distinct lack of usefulness.

I could correlate number of losses (0, 1, maybe 2) with national championship victories (true/false), but how would that help anyone? My recommendation would be: just lose zero games all season and we will win the championship. Wonderful.

Didn't work for UCF.

the correlation between 2 losses and a natty is pretty crappy. So is 1 loss, come to think of it. As pointed out above, 0 wouldn't even correlate that well.

Warning: this post occasionally contains strong language (which may be unsuitable for children), unusual humor (which may be unsuitable for adults), and advanced mathematics (which may be unsuitable for liberal-arts majors)..

username checks out

"Why gobble gobble chumps asks such good questions, I will never know." - TheFifthFuller