2023 Fall Camp | Page 17 | The Boneyard

2023 Fall Camp

huskidork

bailey
Joined
Jul 14, 2023
Messages
133
Reaction Score
347
so so close to the season, I'm excited I expect us to start of well Go Huskies!
 
Joined
Aug 27, 2011
Messages
1,288
Reaction Score
4,933
Despite last season's success.. the computer ratings generally still had the Huskies pretty low on their rankings, with the majority of the wins coming against teams rated in that sub-90 range. Many of the beginning of the year ratings are last year's rating with an adjustment for returning players. Given that it's not a surprise that the computer ratings still don't have a glowing opinion.

I do wonder to what extent the computer ratings consider talent transfers in-bound (although that would probably get even more subjective than considering talent departures).
 
Joined
Sep 18, 2011
Messages
5,047
Reaction Score
19,929
What hurt UConn's computer rankings last year were the game 3 through 5 losses:

Syracuse 48-14
@Michigan 59-0
@NC State 41-10
 
Joined
Aug 26, 2011
Messages
8,194
Reaction Score
15,417
NC ST will have a new QB this year
They have talent on D, some holes to fill on offense. But if they're buying bad press on us or living in the past I think we make them pay. My way too early prediction is we win by more than 7 points. Only key injuries would change that prediction.
 
Joined
Sep 21, 2019
Messages
254
Reaction Score
944
At the beginning of the year computer ranking are largle based on human guesses.
Well thats just not true. A data model is derived from datasets and patterns in said data. The differences in models is what data they use and how they interpret it, meaning if lets say model A heavily weights projected wins on shoelace brand and model B on last 5 seasons win rate, you will likely get very different outputs.

Or even 10 year historical data versus 5 years historical data will show major differences
 

Chin Diesel

Power of Love
Joined
Aug 24, 2011
Messages
32,696
Reaction Score
99,614
Well thats just not true. A data model is derived from datasets and patterns in said data. The differences in models is what data they use and how they interpret it, meaning if lets say model A heavily weights projected wins on shoelace brand and model B on last 5 seasons win rate, you will likely get very different outputs.

Or even 10 year historical data versus 5 years historical data will show major differences

So, they have to make assumptions or guesses on how much to weight different factors?
 
Joined
Aug 28, 2011
Messages
10,640
Reaction Score
16,067
Why make an announcement, is there a rule that there needs to be one on a week before the season?
 
Joined
Sep 21, 2019
Messages
254
Reaction Score
944
So, they have to make assumptions or guesses on how much to weight different factors?
Not even. With proper tools, you can create processes that figure out what data points are relevant or insignificant. Its not at all about football, its just data. The more the better.

Plus wait until this leads to ChatGPT applications that crunch the numbers and help people bet their money more effectively.
 
Joined
Jun 6, 2021
Messages
603
Reaction Score
2,262
Of all the things GPT can do... analytics and math is DEFINITELY not up there
1692593260779.png
 

Online statistics

Members online
143
Guests online
1,083
Total visitors
1,226

Forum statistics

Threads
157,352
Messages
4,096,095
Members
9,984
Latest member
stanfordnyc


Top Bottom