B252 Predictions

[professor ape]

Because (like most of you) I consider mathematical modeling to be a good time, I am playing around with a spreadsheet that predicts standings based on the overall ratings of the active roster. Obviously it is a very rough model but the observable correlation is pretty solid.

With that, based on the Opening Day rosters here are my predictions for the Bronze 252 league which is starting up today (Feb 11th). It’s looking promising for my Market Square Heroes but with the margin of error, I would give myself about a two-in-three shot of winning my division. Nome is close to a lock to win theirs.

AC West Division
Caltech Beavers 94-68 -
Southport Slam Pigs 80-82 14GB
Jerez Sherry Ballers 80-82 14GB
Amsterdam Starters AS 77-85 17GB
Burlington Blue Jays 75-87 19GB

AC Central Division
Market Square Heroes 95-67 -
Conyers Trash Pandas 85-77 10GB
Taichung KMTer 79-83 16GB
Wildwood Prospects 78-84 17GB
Visalia Mercinaries 77-85 18GB

AC East Division
Gogebic Cardinals 82-80 -
Mansfield Buccos 77-85 5GB
Belton Mud Butts 76-86 6GB
Santa Cruz Tweakaz 75-87 7GB
Detroit Maples 75-87 7GB

NC West Division
Nome Gnomes 95-67 -
Madison Cardinals 78-84 17GB
Los Angeles Dodgers 78-84 17GB
Hartford Yard Goats 77-85 18GB
Omaha Sundevils 76-86 19GB

NC Central Division
Sandusky Coasters 92-70 -
Busan Giants 91-71 1GB
Franklin TadPoles 89-73 3GB
Punxatawny Ground Hogs 80-82 12GB
Montreal Expos 79-83 13GB

NC East Division
Columbus Dragons 82-80 -
Bangor Bar Stools 77-85 5GB
Montreal Expos 77-85 5GB
Champaign Colt 45s 77-85 5GB
Heemstede RCH-Silverbacks 74-88 8GB

[stl jason]

Originally Posted by professor ape

Because (like most of you) I consider mathematical modeling to be a good time, I am playing around with a spreadsheet that predicts standings based on the overall ratings of the active roster. Obviously it is a very rough model but the observable correlation is pretty solid.

With that, based on the Opening Day rosters here are my predictions for the Bronze 252 league which is starting up today (Feb 11th). It’s looking promising for my Market Square Heroes but with the margin of error, I would give myself about a two-in-three shot of winning my division. Nome is close to a lock to win theirs.

Marillion fan/inspired, or another source?

[professor ape]

Originally Posted by stl jason

Marillion fan/inspired, or another source?

Absolutely! I’ll be in Montreal in May.

[stl jason]

Originally Posted by professor ape

Absolutely! I’ll be in Montreal in May.

nice... stumbled across them back in the mid 80s... great mix of intricate music/lyrics and some fabulous album art (lots to pick from for a nice logo)

[professor ape]

Originally Posted by stl jason

nice... stumbled across them back in the mid 80s... great mix of intricate music/lyrics and some fabulous album art (lots to pick from for a nice logo)

I got hooked in ‘88 right when they had the transition of lead singer. While always a niche band, they have continued to put out great music. Their most recent two albums are as strong as ever. Check out F.E.A.R. and Sounds That Can’t Be Made if you are looking for something new.

[CrazyWR]

you say the correlation is strong, does this mean you have run this on other teams/seasons and seen that or is this your first season in PT?

[<Pion>]

Sounds like a cool idea, but as much as teams change (especially in lower levels such as this) throughout the season, I would think it would hard to predict. It would be interesting to see the results from a league that is more stable in team make-up such as maybe is found in Perfect (though I would think all leagues change). Maybe once OOTP20 comes out, a league or leagues could be found where very little changes/additions occur.

[professor ape]

Originally Posted by <Pion>

Sounds like a cool idea, but as much as teams change (especially in lower levels such as this) throughout the season, I would think it would hard to predict. It would be interesting to see the results from a league that is more stable in team make-up such as maybe is found in Perfect (though I would think all leagues change). Maybe once OOTP20 comes out, a league or leagues could be found where very little changes/additions occur.

It’s really just a toy at this point to see if there is a reasonable correlation between the “Overall” ratings of the players on teams’ rosters and their chance of success. I wish that I could view leagues other that my own to build up a larger data sample.

[professor ape]

For a model with limited data, this did pretty well in its first full season. It correctly predicted four of the six division winners (including my boys) and was within five wins of the actual totals for 21 of the 30 teams. One division that was missed had predicted the top three to finish within three games of each other. The season ended with those three teams separated by 3.5 games but in a different order. I’ll take that. The one division I did blow was due to a weaker team riding an unexpectedly hot start to the season to finish nine games better than modeled to beat out my predicted winner by five games.
There wasn’t too much change in rosters through the season except for a few teams. One in my division must have spent some money on gold packs because they went from two 80+ players at the start of the season to a whopping 16 in July. As you might expect, they outperformed their predicted number by a ton.
So with a few outliers, the average prediction was 5.5 wins off of the end results which isn’t great but I’m happy with it as a rough modeling toy. I’ll keep playing with it and see if I can bring that down a bit. I’ll be thrilled if the average gap between actual and predicted wins can get down below four games.