iiinova — where your digital self keeps evolving

Bill Radjewski @collegefootballdata.com · 29/09/2026

Two teams can have the same scoring margin and still have performed very differently. My latest tutorial walks through opponent adjustment from first principles, then builds an SRS-style rating in Python and applies it to real college football data. buff.ly/jmHsbef

buff.ly

Raw Stats Lie: Opponent Adjustment Explained

Learn why raw college sports statistics can mistake schedule for skill, then build a simple opponent-adjusted rating you can inspect and improve.

011

Bill Radjewski @collegefootballdata.com · 29/09/2026

Through Week 4, UCLA’s Wayne Knight leads all running backs in opponent-adjusted EPA per rush at 0.49. Memphis’ Jaylin Carter (0.43) and Florida’s Jadan Baugh (0.42) round out the top three. Here are the top 15 most efficient RBs so far:

010

Bill Radjewski @collegefootballdata.com · 28/09/2026

The most exciting game of Week 4? Nevada 33, Air Force 36 - a 9.92 on the CFBD Game Excitement Index. Marshall–Gardner-Webb came in close behind at 9.18. Here are the full top 10:

000

Bill Radjewski @collegefootballdata.com · 28/09/2026

CFBD Model Pick'em - Week 4 Update Yours truly took the top spot on the overall leaderboard this week. Here's how the rest of the Top 10 shook out.

110

Bill Radjewski @collegefootballdata.com · 28/09/2026

North Dakota State’s Drew Klein sits atop the Week 4 PAAR leaderboard at 7.06 Points Added Above Replacement. Michigan State’s Liam Boyd (6.72) and Wisconsin’s Gavin Lahm (6.66) round out the top three. Top 15 kickers through Week 4:

011

Bill Radjewski @collegefootballdata.com · 27/09/2026

Who’s controlling the line of scrimmage through Week 4? This plots opponent-adjusted run blocking vs. opponent-adjusted run defense across FBS. ➡️ Better blocking ⬆️ Better run defense ↗️ Strong on both sides A look at the trenches so far in 2026 👇

110

Bill Radjewski @collegefootballdata.com · 27/09/2026

Yesterday’s 20-19 loss to Iowa checks in at #4 among Michigan’s unlikeliest losses since 2010. Michigan had an 82.8% postgame win expectancy despite coming away with the loss. Here are all 11 games where Michigan lost with a win expectancy above 50%:

2142

Bill Radjewski @collegefootballdata.com · 27/09/2026

Jayden Maiava and Darian Mensah sit atop the Weighted EPA per dropback leaderboard through Week 4, both at 0.38. There’s a four-QB cluster right behind them at 0.32. Here are the top 15 QBs so far in 2026 👇

020

Bill Radjewski @collegefootballdata.com · 27/09/2026

CORE through Week 4 is here. For the first time this season, the ratings are based entirely on 2026 results, no preseason priors. Offense → higher is better Defense → lower is better Upper-right = above average on both sides Here’s the FBS landscape:

182

Bill Radjewski @collegefootballdata.com · 27/09/2026

Great effort from Dylan Raiola coming in for the injured Dante Moore in Oregon's 41-27 win over USC. He led all players with 23.1 EPA. I had Oregon with a 78% postgame win expectancy. Check out the advanced box score for more: buff.ly/5vqmKE9

061

Bill Radjewski @collegefootballdata.com · 27/09/2026

Iowa had a 19% postgame win expectancy in their last-second 20-19 victory over Michigan. Michigan statistically dominated most advanced stats categories, including a 45% to 27% success rate advantage. More from the advanced box score: buff.ly/k92iZt9

060

Bill Radjewski @collegefootballdata.com · 26/09/2026

Midway through the 3rd, Michigan has had 7 completed offensive drives and has been in scoring position on 6 of them. Only averaging 1.5 points per opportunity. Gotta start finishing drives.

030

Bill Radjewski @collegefootballdata.com · 26/09/2026

Michigan leads Iowa 10-7 at half despite dominating statistically. The Wolverines are currently at 75% "Deserve to Win" based on live, in-game advanced metrics.

040

Bill Radjewski @collegefootballdata.com · 26/09/2026

New schedule and game preview pages are now live! The schedule page shows current week games with direct links to Advanced Box Scores and Game Preview pages: buff.ly/dgrbO7f Game Previews break down advanced stats side by side: buff.ly/94m5IcM

030

Bill Radjewski @collegefootballdata.com · 26/09/2026

Our first Friday game is in the books with Army's 21-17 victory over Temple. Per the CFBD Advanced Box Score, Army had a postgame win expectancy of 45% based on the game's advanced stats. More on the advanced box score page: buff.ly/D8VjTF1

030

Bill Radjewski @collegefootballdata.com · 25/09/2026

Only one FBS game Thursday night, which makes Coastal Carolina–Liberty a good chance to dig a little deeper. The revamped CFBD Advanced Box Score includes quarter-by-quarter efficiency splits, scoring and field position, passing-location analysis, havoc breakdowns and much more. 🔗 buff.ly/6n63IY4

010

Bill Radjewski @collegefootballdata.com · 24/09/2026

Another big CFBD website enhancement has gone live this week! Check out the completely revamped Team Trends page (formerly SP+ trends). Explore historical team trends across various ratings and metrics, such as CORE, SRS, Elo, SP+, and FPI. Let me know what you think! buff.ly/HG5lOvh

020

Bill Radjewski @collegefootballdata.com · 24/09/2026

Looking for the best games on the Week 4 slate? CFBD’s CORE index ranks the top FBS matchups using 55% team quality + 45% competitiveness. Top 3: Utah at Iowa State - 92.2 Missouri at Mississippi State - 91.9 Ole Miss at Florida - 88.7 Full top 10 below.

010

Bill Radjewski @collegefootballdata.com · 23/09/2026

Who’s sustaining net efficiency and who’s changing after halftime? Net success rate by half through Week 3: ↗️ Stronger than opponents in both halves ↙️ Weaker in both halves ↖️ Slow start, stronger finish ↘️ Stronger start, weaker finish Dashed lines = FBS medians.

030

Bill Radjewski @collegefootballdata.com · 23/09/2026

Boise State has been almost perfectly balanced running left vs. right this season. The production is anything but balanced: ⬅️ Left: 22 carries • 77.3% success • 13.6 YPC ➡️ Right: 26 carries • 42.3% success • 4.0 YPC A pretty striking directional split.

010

Bill Radjewski @collegefootballdata.com · 22/09/2026

Fan of the revamped advanced box scores? Then check out the completely revamped team profiles on the site! Drill down into ratings, advanced stats, and play location data. Check it out: buff.ly/FoW6oEb Or select a team here: buff.ly/7MTwhMs

110

Bill Radjewski @collegefootballdata.com · 22/09/2026

Weekly 2026 model training data updates are coming soon for Patreon Tier 3+. Want to get your CFB modeling workflow ready before the weekly drops begin? The Model Training Pack includes pre-compiled data and guided notebooks to get you started. buff.ly/QOzw8mr

010

Bill Radjewski @collegefootballdata.com · 21/09/2026

CFBD’s Advanced Box Score just got a complete rebuild. The goal: make it easier to understand how a game was actually played, not just who won. New layout, richer metrics, situational splits, player-level impact, and more. buff.ly/dcen7K9

210

Bill Radjewski @collegefootballdata.com · 21/09/2026

CFBD Model Pick'em - Week 3 Results Top 10 overall models by composite ranking. Congrats to @drawsam.bsky.social and @redbyrd69.bsky.social for both taking the top spot this week!

120

Bill Radjewski @collegefootballdata.com · 21/09/2026

Nolan Hauser sits atop the kicker leaderboard through Week 3. The Clemson kicker has generated 7.58 Points Added Above Replacement (PAAR) on 9 attempts, followed by NDSU’s Drew Klein at 7.06 and Wisconsin’s Gavin Lahm at 6.10. Full top 15 👇

021

Bill Radjewski @collegefootballdata.com · 20/09/2026

Vanderbilt-NC State was the game to watch in Week 3. Vandy’s 35-31 win posted an 8.99 Game Excitement Index, followed by Toledo-Temple at 7.73 and Louisville-SMU at 7.08. The full Week 3 top 10 👇

030

Bill Radjewski @collegefootballdata.com · 20/09/2026

Week 3 brought 11 CORE upsets across 75 games. Kentucky was the biggest shocker, entering with just an 8.5% win chance against Texas A&M. SE Louisiana (18.6%), West Virginia (29.9%), and Ole Miss (31.6%) weren’t far behind. Full rundown 👇

011

Bill Radjewski @collegefootballdata.com · 20/09/2026

The early two-way contenders are starting to separate. Alabama, Notre Dame, and Ohio State all sit firmly in the “stronger on both sides” quadrant through Week 3. And then there’s that Mississippi State offense. 👀 25% preseason / 75% observed blend.

141

Bill Radjewski @collegefootballdata.com · 17/09/2026

The CFBD Week 3 watchlist is set. Top matchups by Watchability Index: Mississippi State at South Carolina - 94.5 Florida at Auburn - 93.3 Miami at Wake Forest - 89.9 LSU at Ole Miss - 88.1 SMU at Louisville - 86.0 Plenty worth watching this weekend.

021

Bill Radjewski @collegefootballdata.com · 14/09/2026

CFBD Model Pick'em - Week 2 Results Top 10 overall models by composite ranking. @ArnoldJohn_cfb takes the top spot!

120

Bill Radjewski @collegefootballdata.com · 13/09/2026

The most exciting games of Week 2, according to the CFBD Game Excitement Index. Purdue-Wake Forest leads the way at 10.16, followed by Ohio-Jacksonville State (9.76) and Oregon State-Texas Tech (8.80). Plenty of close finishes and late-game tension this week.

010

Bill Radjewski @collegefootballdata.com · 13/09/2026

The Week 2 CFBD CORE landscape. USC has separated itself offensively, while LSU and Alabama lead a loaded group of defenses. The model is also becoming much more results-driven: CORE is now 50% preseason / 50% observed performance.

041

Bill Radjewski @collegefootballdata.com · 13/09/2026

Ohio State had a 70% postgame win expectancy in their 23-24 loss to Texas. Texas had an 83% (!) success rate on passing downs in the 4th quarter. Advanced Box Score - buff.ly/it0GlO6 Win probability chart below.

1130

Bill Radjewski @collegefootballdata.com · 12/09/2026

CORE pregame favorites are 23–2 for Week 2 through the games completed so far. The two upsets, with each winner’s pregame CORE win chance: Oklahoma State over Oregon: 11.6% App State over East Carolina: 29.0%

010

Bill Radjewski @collegefootballdata.com · 12/09/2026

Oregon had a 76% postgame win expectancy in a 31-39 LOSS to Oklahoma State. Oregon rated higher in EPA/play and explosiveness, but had a 37% success rate on standard downs while allowing 51% success to Oregon State in such situations. Advanced Box Score - buff.ly/6MCwzQa

collegefootballdata.com

CollegeFootballData.com

020

Bill Radjewski @collegefootballdata.com · 12/09/2026

I had Michigan with a 84% postgame win expectancy in a 17-10 win over Oklahoma. Story of the game: success rate on passing downs. I had the Wolverine offense at 40% success in such situations compared to 25% for OU. Advanced Box Score - buff.ly/0j9Sl6V

collegefootballdata.com

CollegeFootballData.com

0322

Bill Radjewski @collegefootballdata.com · 12/09/2026

Week 2 is loaded. 👀 Ohio State-Texas leads this week’s CFBD Watchability rankings at 94.4, with Oklahoma-Michigan close behind at 91.8. Here are the 10 FBS matchups scoring highest on our watchability index heading into Saturday. 🏈

010

Bill Radjewski @collegefootballdata.com · 09/09/2026

Here’s the Week 1 CFBD CORE offense vs. defense landscape. The upper-right quadrant is where you want to be: stronger than FBS average on both sides of the ball. Ohio State, Georgia, Notre Dame, Oregon, and USC are among the early standouts. Current blend: 75% preseason / 25% observed.

041

Bill Radjewski @collegefootballdata.com · 08/09/2026

Week 1 delivered. Charlotte-The Citadel tops the CFBD Game Excitement Index at 9.44, followed by Ole Miss-Louisville (8.77) and Buffalo-UAlbany (7.84). Here are the 10 most exciting games from the opening week of the 2026 season.

010

Bill Radjewski @collegefootballdata.com · 08/09/2026

Preseason expectations got tested immediately. Across 99 games from Aug. 29–Sept. 7, there were 10 underdog wins. The biggest preseason upsets by CFBD CORE win chance were Idaho State over Utah State (5.5%), UMass over Rutgers (8.8%), and Tarleton State over Bowling Green (10.9%).

000

Bill Radjewski @collegefootballdata.com · 08/09/2026

The first CFBD CORE Ratings of 2026 are in. Ohio State leads the way through Week 1, followed by Georgia and Notre Dame. Texas Tech jumps to #4 on the strength of an elite defensive rating. Still early, but we're starting to get our first signal from the 2026 season.

000

Bill Radjewski @collegefootballdata.com · 08/09/2026

CFBD Model Pick'em - Week 1 Results Top 10 overall models by composite ranking. @mrstatsnstuff takes the top spot! Lots of great models this year. Feels like every year this thing just gets more and more competitive.

120

Reposted by Bill Radjewski

Darian Somers @stuffsomerssays.com · 06/09/2026

I know it's something you can get from a lot of national outlets or on your phone but I really am tickled about having box scores on the blog again. Wouldn't be possible without @collegefootballdata.com stuffsomerssays.com/psu-football...

stuffsomerssays.com

Box Score: Penn State 45, Marshall 0 - Sept. 5, 2026

Penn State-Marshall Box Score

022

Bill Radjewski @collegefootballdata.com · 05/09/2026

CORE’s preseason favorites are 22–5 through Sept. 4. The five upsets and each winner’s win probability: UMass over Rutgers: 8.8% Colorado over Georgia Tech: 25.0% UNC over TCU: 35.0% Michigan State over Toledo: 40.7% San José State over EMU: 48.8% CORE expected 5.05 upsets. We got 5.

000

Bill Radjewski @collegefootballdata.com · 05/09/2026

Happy College Football Saturday. Model Pick'em leaderboard is live for 2026! Be sure to get your picks in ahead of today's action. And follow along live at buff.ly/W0tSmrj.

030

Bill Radjewski @collegefootballdata.com · 03/09/2026

Five new CFBD rushing endpoints are live. Analyze PPA, success rate, line yards, stuff rate, power success, explosiveness, and more by direction: left, middle, or right. Coverage: partial in 2025; mostly full from 2026 onward. api.collegefootballdata.com/api/rushing

020

Bill Radjewski @collegefootballdata.com · 02/09/2026

I’m often asked how to get started building with college football data. The CFBD Gumroad packs are designed as an on-ramp: pre-compiled data + guided Jupyter notebooks in the Starter/Model Training packs, and AI + CFBD fundamentals in the AI packs. buff.ly/qwLi5yU

030

Reposted by Bill Radjewski

Brian Fremeau @bcfremeau.bsky.social · 02/09/2026

I wrote a little bit about how I compile data in my possession efficiency ratings work, and how I rely on a variety of sources and friends (from @collegefootballdata.com to @sickoscommittee.org, if necessary) to build a data set and clean it up: bcftoys.com/whiteboard/s...

bcftoys.com

sources — BCF Toys

I’m asked from time to time about how I gather the data to produce FEI ratings and other metrics I publish on this site. I utilize several sources, but a reliable account of the results of every posse...

1165

Reposted by Bill Radjewski

Brian Fremeau @bcfremeau.bsky.social · 30/08/2026

I can't overstate my appreciation for folks like Bill Radjewski @collegefootballdata.com making a wealth of data available and accessible for people to make cool things and uncover analytics insights: collegefootballdata.com

collegefootballdata.com

2026 CFB Analytics Packs, College Football API & Tools

Get the 2026 Model Training Pack, CFB Analytics Bundle, CFB Season Builder Bundle, Starter Pack, college football API access, data exports, and analytics tools.

1154