Realignment stretched the conferences
Average distance between member stadiums, 2019 vs 2024–26 (schools in this dataset only). Longer bars mean longer road trips — and less regional identity.
1st Quarter
Where the songs took root
Every dot on this map is a stadium, and every dot is beating at the tempo of its fight song. Scroll to see how conferences, realignment and rivalries carve up the country.
65 heartbeats
Each stadium pulses at its song's tempo. Lighter, quicker dots are faster songs. Even before any analysis, you can see there is no obvious “fast region” of the country.
Click any dot to open the school's card and hear the song.
The 2019 map
In 2019, when FiveThirtyEight collected this data, the five power conferences were tidy regional blocs. The shaded shapes are each conference's footprint (its convex hull).
Big Ten members were on average — apart.
2024: the map is redrawn
Realignment pulled USC, UCLA, Oregon and Washington into the Big Ten, Texas and Oklahoma into the SEC, and Stanford and Cal into the Atlantic Coast Conference. Only Oregon State and Washington State were left in the Pac-12.
Average Big Ten distance: — → —. The ACC: — → —.
Every song has a turf
Split the country by whichever stadium is closest and you get each song's “home turf” (a Voronoi diagram). Drop a pin anywhere and you know which fight song you'd hear on Saturday.
Try .
Lines of hostility
33 classic rivalries connect these schools. Most are short: the average rival is — away, versus — for a random pair. Rivalry is mostly a neighbourhood affair.
Your turn
Use the toolbar to recolor the map, switch eras, or toggle layers. Click a conference in the legend to filter every chart on the page.
Does geography shape the sound?
We tested whether nearby schools have similar songs using Moran's I (each school vs its 5 nearest neighbours, 9,999 random shuffles). Grey histograms show what “no geographic pattern” looks like; the line is the real data.
2nd Quarter
A golden age, 1905–1925
Just over half of all dated fight songs were written in two decades, as college football exploded in popularity. Press play to watch them appear across the map.
When each song was written
— songs have no known year and are not shown.
Spreading across the map
1970
The gold trail follows the centre of gravity of every song written so far, decade by decade.
Halftime
Two tempos — or one?
The raw data shows two clusters of tempo: around 75 BPM and around 150 BPM. But 75 is exactly half of 150. Marching band music is often written in cut time, so the same pulse can be counted either way.
Tempo (beats per minute)
Folding doubles every song under 100 BPM (— songs). The bimodality coefficient drops from — to — (above 0.555 suggests two peaks). Once the counting is aligned, nearly every fight song lives in the same 130–180 BPM march range.
Song length (seconds)
Most songs run 40–100 seconds. The long tail is repeats: Aggie War Hymn cycles the same melody several times.
Energy vs. stamina
Tempo against length. The quadrants are a playful guide, not a statistical grouping.
Tap the tempo
Tap the pad (or press Space while it's focused) to a beat in your head. We'll find the fight songs that march at your speed.
—BPM
3rd Quarter
Fight, win, rah — repeat
FiveThirtyEight tagged eight lyrical clichés in every song. Some are nearly universal; others mark out a school's personality.
Who uses which cliché?
Each dot is a song that uses the trope, grouped by conference. Hover for the school, click to open its card.
Conference signatures
Which clichés travel together?
Correlation (φ) between tropes. appear together more than chance; tend to exclude each other.
The “fightiest” songs
Each football is one shout of “fight”. Texas Fight is in a league of its own.
Written by students?
One square per song. Students wrote about half of them, a reminder that many traditions started in the stands, not the president's office.
Four kinds of fight song
Clustering songs by tempo, length, era, authorship and all eight tropes (Gower distance, Ward linkage) gives four archetypes. Songs that sit close together in this “galaxy” are musically similar.
4th Quarter
Face-off
Do rivals sound alike, or do they deliberately sound different? Pick a rivalry and put the two songs head to head.
33 rivalries
Head to head
Near neighbours, different songs
Every pair of schools (— pairs): distance apart vs how different their songs are. If neighbours copied each other, points would rise to the right. They don't (Mantel r = —, p = —). Rivals, in gold, are no more alike than random pairs (permutation p = —).
Local hot and cold spots
Local Moran's I (LISA) flags schools whose neighbourhood is unusually high or low. With 65 schools and many tests, treat these as hints, not findings.
Overtime
Where's that fight song?
Five songs, no names. Read the clues, listen if you like, then drop a pin where you think the stadium is. The closer you are, the more points.
Post-game
Methods & caveats
Data
- Fight song metadata and lyric tropes: FiveThirtyEight, with BTAA updates. Stadium coordinates: CFBD API.
- The BTAA update mixes years (some schools already in their 2024 conference, others not), so conference membership was re-coded for 2019 and 2024–26.
- Rivalries are a curated list of 33 long-running series between schools in the dataset.
Analysis
- Folded tempo: songs under 100 BPM are doubled, a common correction for half-time counting.
- Musical distance: Gower distance over folded BPM (weight 2), length, year, student authorship and eight tropes.
- Spatial statistics: Moran's I and LISA with 5-nearest-neighbour weights; Mantel test between great-circle and musical distance. All p-values come from 9,999 permutations (999 for LISA).
- Conference differences: permutation χ² test; intervals are Wilson 95%.
Caveats
- 65 songs is a small sample. Several conferences have fewer than 15 schools, so a single song moves a percentage by 7+ points.
- Eight tests on conference differences means one “significant” result is not surprising by chance alone.
- Reproduce everything with
python analysis/build_data.py.