Research

We publish what we tested

Popular sports ideas — hot hands, short weeks, home field — tested against our own game logs. We publish every result, including the ones that found nothing.

Live studies

Recomputed from our game logs · every player measured against their own baseline

NBATeams lose more

NBA back-to-backs: teams feel it, scorers don't

Do teams lose more on the second night of a back-to-back — and do their players score less?

46.3%
team win rate on back-to-backs (vs 50.9% rested)
Back-to-backRested
Team win rate46.3%50.9%
Player pts vs own avg+0.11−0.01
Games9024,000

Teams won 46.3% of back-to-back second nights and 50.9% otherwise (z = -2.5). But players who suited up scored +0.11 points vs their own average (z = 1.4) — no drop. The likely reason is load management: the most tired players sit, so they never appear in the back-to-back sample at all.

How we measured it · 36,382 player-games · 4,902 team-games▾

NBA 2024-25 and 2025-26 and 2026-27 regular seasons. Back-to-back = team played the previous day (from each team's game dates). Player rows: 15+ minutes, players with 20+ such games, scored against their own season average.

Market backtests

Already live · deepest example today

What We've Measured — College Football

Four angles we tested and killed on our own archive of FBS results and closing lines — SP+ ratings vs the spread (49.8% ATS on 1,589 games), wind, rain and altitude on totals. Published with the same numbers whether they worked or not.

Read the full page →

Other written studies

Our track record

Descriptive market statistics, computed daily

64%

of MLB props priced +200 or longer

Of 1,241,776 graded props this season, 790,940 closed at +200 or longer. Staked flat at one unit each, that group returned -29.9%. This describes what the shelf has done, not a strategy.

mlb · recomputed daily from our odds archive

CFB is our deepest write-up because that market has been backtested most extensively — but it's not the only one. NFL and WNBA have their own published number studies above, and our full pick history is public too. Same standard everywhere: publish it whether it worked or not.