Backtest History

3859 graded games

Every graded prediction from the frozen walk-forward backtest, next to the closing spread and the final margin. Backtest seasons are historical stand-ins, not live results. Lines are quoted for the home team.

CFB backtest history
Season	Wk	Matchup	Score	Model	Close	Result	Model err
2023	7	UNLV @ Nevada	45–27	+10.3	+7.5	+18.0	7.7 (model closer than the closing line)
2023	7	Wyoming @ Air Force	27–34	-18.7	-12.5	-7.0	11.7
2023	6	Jacksonville State @ Middle Tennessee	45–30	+4.1	-3.0	+15.0	10.9 (model closer than the closing line)
2023	6	Western Kentucky @ Louisiana Tech	35–28	+4.7	+6.0	+7.0	2.3
2023	6	Sam Houston @ Liberty	16–21	-24.8	-21.0	-5.0	19.8
2023	6	Florida International @ New Mexico State	17–34	-9.7	-6.5	-17.0	7.3 (model closer than the closing line)
2023	6	Boston College @ Army	27–24	-9.0	-2.5	+3.0	12.0
2023	6	Vanderbilt @ Florida	14–38	-17.8	-18.5	-24.0	6.2
2023	6	Kentucky @ Georgia	13–51	-8.4	-14.5	-38.0	29.6
2023	6	Nebraska @ Illinois	20–7	-0.7	-3.0	+13.0	13.7 (model closer than the closing line)
2023	6	Purdue @ Iowa	14–20	+0.5	-2.5	-6.0	6.5
2023	6	Michigan @ Minnesota	52–10	+21.8	+18.5	+42.0	20.2 (model closer than the closing line)
2023	6	Western Michigan @ Mississippi State	28–41	-10.4	-21.0	-13.0	2.6 (model closer than the closing line)
2023	6	LSU @ Missouri	49–39	+6.4	+6.0	+10.0	3.6 (model closer than the closing line)
2023	6	FCS @ Northwestern	20–23	-25.1	-23.0	-3.0	22.1
2023	6	Maryland @ Ohio State	17–37	-10.7	-17.0	-20.0	9.3
2023	6	Arkansas @ Ole Miss	20–27	-10.7	-13.0	-7.0	3.7 (model closer than the closing line)
2023	6	UConn @ Rice	38–31	-13.5	-10.0	+7.0	20.5
2023	6	Alabama @ Texas A&M	26–20	-3.5	+2.5	+6.0	9.5
2023	6	Rutgers @ Wisconsin	13–24	-10.3	-13.0	-11.0	0.7 (model closer than the closing line)
2023	6	Arizona @ USC	41–43	-7.5	-21.0	-2.0	5.5 (model closer than the closing line)
2023	6	Colorado @ Arizona State	27–24	+4.5	+3.0	+3.0	1.5
2023	6	Oregon State @ California	52–40	+21.1	+7.5	+12.0	9.1
2023	6	Washington State @ UCLA	17–25	-1.0	-3.5	-8.0	7.0
2023	6	Notre Dame @ Louisville	20–33	+5.3	+6.0	-13.0	18.3 (model closer than the closing line)
2023	6	Wake Forest @ Clemson	12–17	-29.0	-21.0	-5.0	24.0
2023	6	Virginia Tech @ Florida State	17–39	-19.0	-24.0	-22.0	3.0
2023	6	Georgia Tech @ Miami	23–20	-23.2	-19.0	+3.0	26.2
2023	6	Syracuse @ North Carolina	7–40	-8.1	-9.5	-33.0	24.9
2023	6	Marshall @ NC State	41–48	-7.1	-6.5	-7.0	0.1 (model closer than the closing line)
2023	6	FCS @ Virginia	13–27	-23.6	-10.0	-14.0	9.6
2023	6	Kansas State @ Oklahoma State	21–29	+16.2	+11.5	-8.0	24.2
2023	6	Texas Tech @ Baylor	39–14	+7.4	+2.5	+25.0	17.6 (model closer than the closing line)
2023	6	TCU @ Iowa State	14–27	+13.6	+6.5	-13.0	26.6
2023	6	UCF @ Kansas	22–51	-4.6	+2.0	-29.0	24.4 (model closer than the closing line)
2023	6	Oklahoma vs Texas	34–30	-0.2	-4.0	+4.0	4.2 (model closer than the closing line)
2023	6	Toledo @ Massachusetts	41–24	+16.9	+19.0	+17.0	0.1 (model closer than the closing line)
2023	6	Tulsa @ Florida Atlantic	17–20	+4.3	-3.0	-3.0	7.3
2023	6	South Florida @ UAB	35–56	+9.7	+3.0	-21.0	30.7
2023	6	UTSA @ Temple	49–34	+8.6	+14.0	+15.0	6.4
2023	6	North Texas @ Navy	24–27	-1.5	-6.5	-3.0	1.5 (model closer than the closing line)
2023	6	Texas State @ Louisiana	30–34	-0.5	-1.0	-4.0	3.5
2023	6	South Alabama @ UL Monroe	55–7	+5.5	+11.0	+48.0	42.5
2023	6	Old Dominion @ Southern Miss	17–13	-1.8	-3.0	+4.0	5.8 (model closer than the closing line)
2023	6	Arkansas State @ Troy	3–37	-17.7	-15.5	-34.0	16.3 (model closer than the closing line)
2023	6	Ball State @ Eastern Michigan	10–24	-6.2	-2.5	-14.0	7.8 (model closer than the closing line)
2023	6	Central Michigan @ Buffalo	13–37	-1.7	+2.5	-24.0	22.3 (model closer than the closing line)
2023	6	Bowling Green @ Miami (OH)	0–27	-8.9	-7.5	-27.0	18.1 (model closer than the closing line)
2023	6	Kent State @ Ohio	17–42	-27.7	-24.5	-25.0	2.7
2023	6	Northern Illinois @ Akron	55–14	-2.5	+4.0	+41.0	43.5

Model and Close are home lines; Result is the final margin on the same axis. Model err is the absolute miss of the model's margin, shown green when the model was closer than the closing line.