Brian Fremeau

Starting field position differential is an interesting "core" metric. I've found that it does have a positive relationship with team strength, but not nearly as strong as offensive and defensive efficiency metrics do.

Interesting to me that 7 of 12 core metrics are "raw" unadjusted statistics and five are "opponent-adjusted" (as in, calculated relative to opponent averages)

But I also have more questions!
- Relative scoring offense and scoring defense; I'm curious if these metrics are calculated with *all* game points scored, or only those scored on offensive drives
- Same with points per possession (off/def), yards per point, and plays per point

Love digging into these articles every year on the committee selection process, and this one has some intriguing details about the 12 metrics used as the "bedrock of team comparisons".

FEI games of the year to date (based on the average of each team's opp-adjusted game ratings):

Alabama 24, Georgia 21
Ohio St 14, Texas 7
Oregon 30, Penn St 24
Georgia 44, Tennessee 41
Texas A&M 41, Notre Dame 40
Miami 27, Notre Dame 24
Ole Miss 24, LSU 19
Indiana 20, Iowa 15

"Best losses" to date:
Tennessee vs Georgia (41-44)
Texas vs Ohio St (7-14)
Georgia vs Alabama (21-24)
Penn St vs Oregon (24-30)
ND vs Texas A&M (40-41)
ND vs Miami (24-27)
LSU vs Ole Miss (19-24)
Kentucky vs Ole Miss (19-24)
Iowa vs Indiana (15-20)
Nebraska vs Michigan (27-30)

0% weight when a team has played 10 FBS opponents, or when we reach the end of the regular season, whichever comes first.

90th+ percentile FEI game ratings vs FBS opponents through Week 6:

4x - Alabama
3x - Miami, Notre Dame, Ohio State, Oregon
2x - Florida State, Georgia, Illinois, Indiana, Oklahoma, Old Dominion, Ole Miss, Texas A&M, Texas Tech, Utah, Vanderbilt

bcftoys.com/2025-gr

Weekly ratings progression for every team: bcftoys.com/2025-wfei

Screenshot of FEI ratings table of top-15 teams through Week 6

FEI ratings and supporting metrics are updated through Week 6 results: bcftoys.com/2025-fei
Preseason projections carry 43% weight for teams that have played four FBS opponents, 33% for teams that have played five, 25% for teams that have played six.

FEI ratings aren’t updated yet, they post each week by Tuesday morning

Averaging two teams' GR percentiles in each game as a way to rank the year's best so far...

.913 Ohio St 14, Texas 7
.910 Oregon 30, Penn St 24 (2OT)
.899 Alabama 24, Georgia 21
.868 Georgia 44, Tennessee 41 (OT)

Both teams 80th+ percentile GR in each. Passes a sniff test?

Pretty decent women's hoops Final 4 bracket you got there

Found in my parents' basement. #dataviz from the 1932 Olympics

What I'm sure I can do better at is bridging the gap for those encountering "analytics" they don't immediately understand. There have to be better entry points than just "look at this list of teams, now love/hate/share it".

I *think* I do a fairly decent job at this stuff with FEI but please let me know if I’m missing any marks:

- ratings are clear about what is being measured
- supporting data is there to explain why some ratings might not make intuitive sense

Wins above what an average FBS team would expect to have against the schedule faced through Week 5, per FEI:
+2.9 Ole Miss
+2.4 Texas A&M
+2.3 Ohio State
+2.2 Iowa State
+2.2 Miami
+2.2 Indiana
+1.9 Oklahoma
+1.8 Missouri
+1.8 Arizona State
+1.8 Georgia Tech

FEI without preseason weights are still pretty wonky at this point because of limited data. Oregon would be 10th, for instance, having just defeated #29 Penn State. Top 5 would be:
1. Indiana
2. Miami
3. Notre Dame
4. Texas Tech
5. Alabama

90th+ percentile FEI game ratings (opp-adjusted single game possession efficiency) vs FBS opponents through Week 5:
4 - Oregon
3 - Alabama, Miami
2 - Florida State, Notre Dame, Ohio State, Ole Miss, Texas Tech, Utah, Vanderbilt

Game Ratings: bcftoys.com/2025-gr

Weekly ratings progression for every team: bcftoys.com/2025-wfei

Screenshot of FEI top 15 team rankings table

FEI ratings and supporting metrics are updated through Week 5 results: bcftoys.com/2025-fei
Preseason projections carry 54% weight for teams that have played only three FBS opponents, 43% for teams that have played four, 33% for teams that have played five.

You’re probably also thinking of @bcsknowhow.bsky.social

Yes I’ve tested a number of different weights for the priors.

Preseason FEI projections are based only on five years of priors, no roster adjustments. End of year FEI ratings are exclusively based on in-season data, no priors.

Scatterplots illustrating how weekly team ratings transition from data based entirely on preseason priors to data based entirely on in-season results. Week 4 is highlighted.

We are here.

I had an idea yesterday about making a word search that, when solved, would be a data visualization.

So, I coded it up... and it worked!

This shows abundance of three bird species in Brooklyn Bridge Park over a year.

It feels like there are some fun possibilities here.

📊🪶

If no preseason projection weight was applied:
1. Indiana
2. Florida State
3. Alabama
4. Vanderbilt
5. Miami
6. Texas Tech
7. Notre Dame
8. Old Dominion
9. Oklahoma
10. Mississippi State

Multiple 90th+ percentile FEI game ratings (opponent-adjusted single game efficiency performances) vs FBS opponents through Week 4 :

Alabama
Florida State
Miami
Ole Miss
Oregon (3x)
Vanderbilt

Game ratings: bcftoys.com/2025-gr

Top 25 F+ teams that haven't played a top-60 opponent yet; each will face their first this week:
#1 Oregon (next vs #5)
#5 Penn State (vs #1)
#12 USC (vs #35)
#23 BYU (vs #58)
#25 Washington (vs #2)
bcftoys.com/2025-fplus

Weekly FEI ratings progression for every team: bcftoys.com/2025-wfei