Pick Sheet — 2026 Week 6

Enter the confidence number next to each game. Highest number = most confident. Strategy: Always-favourite (baseline).

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Conf	Pick	Game	Line
14	LA	ARI @ LA	-12.5
13 was 12 ▲1	KC	LAC @ KC	-8.5
12 was 13 ▼1	SF	WAS @ SF	-7.5
11	NE	NYJ @ NE	-7.5
10	IND	TEN @ IND	-7.0
9 was 7 ▲2	BAL	BAL @ CLE	-5.5
8 was 9 ▼1	BUF	BUF @ LV	-5.5
7 was 3 ▲4	SEA	SEA @ DEN	-2.5
6 was 8 ▼2	JAX	HOU @ JAX	-2.5
5 was 6 ▼1	PIT	PIT @ TB	-2.5
4 was 5 ▼1	GB opened on DAL −2.5	DAL @ GB	-1.5
3 was 2 ▲1	CHI	CHI @ ATL	-1.5
2 was 4 ▼2	CAR opened on PHI −2.5	CAR @ PHI	-1.5
1	NO FLIP CANDIDATE	NO @ NYG	-1.5

Archived 2026-10-15 13:00 UTC. These are the picks as made and they don't change. Odds as of 2026-10-15 12:15 UTC. Line shown is the pick's own spread. 10 marked with their opening value because the pick or confidence changed since the opening line. Shaded rows moved 3+ slots or flipped. Opening line: first DraftKings line captured, 2026-10-07 00:15 UTC.

Thursday game locked and lines moved since? Re-sort the rest by live odds →

Optional flip: NO @ NYG (1.5) is closest to a tossup. Taking NYG at confidence 1 costs at most 1.5pp of win probability if rivals pick independently. It gains +36.5pp if they all take the favourites and the pool would otherwise tie. It is a standings move, not a football opinion, and the only deviation with evidence behind it.

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. Highlighted rows are spreads on this week's sheet.

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

Highest scoring team

guess 44 points · implied 32.0

right 13% of weeks (random 3%)

Lowest scoring team

LAC

guess 4 points · implied 17.0

right 18% of weeks (random 3%)

Sunday night · DAL @ GB

DAL 24 – GB 22

total 46 · posted 53.5

Monday night · WAS @ SF

WAS 18 – SF 26

total 44 · posted 51.5

Scores add up to the posted total minus 7, rounded down, and split the way the spread splits it.

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.