Cosme Salas

Expected points added: how much the player’s plays moved the team’s expected score, per play and in total. Plays: total EPA divided by EPA per play, the plays the source credits to him (for a receiver roughly his catches and runs, not every target; for a passer, not every attempt), so it can differ from attempts plus carries. Percentile: against the position’s regulars that season across all divisions on file (quarterbacks with 150+ plays, backs 60+, receivers and tight ends 30+), with how many there were (n) and their median; a dash means too few plays to rank. A season in progress uses the full-season bar scaled by games played, never below 75 plays for quarterbacks, 30 for backs and 20 for receivers. The cohort grows or shrinks with the source’s coverage each season (2025 lists far more players), so compare percentiles rather than raw EPA across years. FCS seasons are marked. Usage: the share of the team’s plays the player was part of (shown from 20 plays). PAAR: points a kicker added above a replacement-level kicker, with his field-goal attempts. From CollegeFootballData, 2013 on.

2026 Duke 4 games, 4 of 5 field goals

Date	Opponent	Result	FGM	FGA	Long	XP	Pts
Sep 5	vs.Tulane	W 17–⁠3	1	2	22	2	5
Sep 12	atIllinois	W 31–⁠27	1	1	39	4	7
Sep 19	vs.Stanford	W 35–⁠7	0	0	0	5	5
Sep 26	vs.William & Mary	W 62–⁠7	2	2	39	5	11

2025 Duke 3 games, 1 of 1 extra point

Date	Opponent	Result	FGM	FGA	Long	XP	Pts
Nov 29	vs.Wake Forest	W 49–⁠32	–	–	–	–	–
Dec 6	vs.Virginia	W 27–⁠20	0	0	0	1	1
Dec 31 ✦	vs.Arizona State	W 42–⁠39	–	–	–	–	–

Box scores from CollegeFootballData, 2004 on. ✦ postseason; ★ a career high on file.