Pick Sheet — 2026 Week 21
Enter the confidence number next to each game. Highest number = most confident. Strategy: Always-favourite (baseline).
No games found for 2026 Week 21.
How often the favourite loses outright
Every 1999–2026 regular-season and playoff game with a closing spread: favourites lost 33.4% of 7,366.
| Fav by | Games | Lost | Loss % | 95% range | Since 2015 | |
|---|---|---|---|---|---|---|
| 1 | 546 | 266 | 48.7% | 45–53% | 51.9% (206) | |
| 1.5 | 218 | 103 | 47.2% | 41–54% | 45.7% (116) | |
| 2 | 169 | 71 | 42.0% | 35–50% | 43.8% (80) | |
| 2.5 | 546 | 254 | 46.5% | 42–51% | 46.0% (326) | |
| 3 | 1,176 | 490 | 41.7% | 39–45% | 39.5% (456) | |
| 3.5 | 702 | 254 | 36.2% | 33–40% | 36.3% (328) | |
| 4 | 342 | 116 | 33.9% | 29–39% | 44.4% (133) | |
| 4.5 | 277 | 94 | 33.9% | 29–40% | 34.2% (114) | |
| 5 | 165 | 49 | 29.7% | 23–37% | 28.6% (63) | |
| 5.5 | 291 | 99 | 34.0% | 29–40% | 34.5% (145) | |
| 6 | 357 | 112 | 31.4% | 27–36% | 29.7% (145) | |
| 6.5 | 347 | 98 | 28.2% | 24–33% | 21.7% (143) | |
| 7 | 500 | 119 | 23.8% | 20–28% | 23.8% (181) | |
| 7.5 | 296 | 69 | 23.3% | 19–28% | 23.4% (171) | |
| 8 | 100 | 29 | 29.0% | 21–39% | 26.8% (41) | |
| 8.5 | 107 | 29 | 27.1% | 20–36% | 31.5% (54) | |
| 9 | 153 | 37 | 24.2% | 18–32% | 31.2% (16) | |
| 9.5 | 163 | 34 | 20.9% | 15–28% | 16.3% (49) | |
| 10 | 217 | 41 | 18.9% | 14–25% | 19.0% (84) | |
| 10.5 | 136 | 22 | 16.2% | 11–23% | 14.1% (64) | |
| 11 | 80 | 14 | 17.5% | 11–27% | 30.8% (26) | |
| 11.5 | 43 | 10 | 23.3% | 13–38% | 18.8% (16) | |
| 12 | 33 | 8 | 24.2% | 13–41% | 14.3% (7) | |
| 12.5 | 49 | 6 | 12.2% | 6–24% | 8.3% (24) | |
| 13 | 73 | 8 | 11.0% | 6–20% | 7.1% (28) | |
| 13.5 | 65 | 10 | 15.4% | 9–26% | 10.7% (28) | |
| 14+ | 215 | 19 | 8.8% | 6–13% | 9.9% (101) |
Ties count as not lost. Pick'em lines are left out. Rows past 8 points have a few dozen to 200 games, so read neighbouring rows together. Rates haven't changed since 2015 (spread-adjusted difference −0.05 points, 95% range −1.8 to +1.8).
Tiebreakers
No Sunday or Monday night game on this week's schedule.
Why the guess is under the line
The posted total is the best forecast available. A tiebreaker rewards being closest, though, and if everyone writes down the line, the line is the most crowded number. Guessing 7 points off it turns a 1% win rate into about 30% in a 20-person pool.
The guess goes under because the median total lands half a point below the line, even though the average is above it.
The team picks usually miss. The highest-implied team tops the week 13% of the time and the lowest-implied bottoms it 18%: four to six times better than guessing. Adding scoring volatility made both picks worse.
The right distance assumes the pool guesses near the line. If it scatters, the distance should grow.
Study: backtests/pickem_tiebreak/FINDINGS.md.
Why the sheet just follows the favourites
Across 2,119 games (2018–2025), always picking the favourite, most confident on the biggest spreads, scored 72.39% of maximum points and went 66.4% straight up. Five alternatives were tested and none beat it. Ranking by devigged moneyline tied it (+0.00pp) while ordering 138 of 141 weeks differently. Flipping small home underdogs was 0.52pp worse (99% interval excludes zero).
The usual way to lose a season-long pool is a handful of gut overrides on close games, where the market is hardest to beat.
The one exception is the optional flip above. It is about standings, not football: one deviation at the bottom of the sheet keeps you from tying a field that all picked the same favourites. One game, never more.