Week 3 runs Thursday, Sep 24 through Monday, Sep 28 — 16 games, lineups lock Thursday, Sep 24 at 8:15 PM ET, and the model has projections on 349 rostered players across the full slate. What those numbers are: per-player DraftKings-point projections built from opportunity volume and per-opportunity production, shrunk for opponent context and pace, expressed as a floor/ceiling band (p10/p50/p90 of a 5,000-draw Monte Carlo per player) — the model is walk-forward validated out of sample, replaying the 2025 season week-by-week from prior data only and beating season-mean MAE and rank correlation on both benchmarks. One honest caveat to carry in: any draftable with zero or near-zero NFL game history in the play-by-play store — rookies and first-year players — carries no projection, because the model has nothing real to work from; those players are omitted rather than assigned invented numbers, and that historically covers 12–15% of Week 1 usage players, so a cheap name with no Arcline number attached is a guess, not a projection. This column is projections and lineup construction, full stop — that's the entire exercise here.

The model's highest-projection players this week, with their DraftKings salaries, projected DraftKings points, floor/ceiling bands, and value per $1K of salary:

• Josh Allen (QB, BUF) — $8,000 | 26.5 proj | floor 9.9 / ceiling 44.8 | 3.31 per $1K
• Jaxon Smith-Njigba (WR, SEA) — $8,600 | 24.5 proj | floor 8.7 / ceiling 41.9 | 2.85 per $1K
• Jahmyr Gibbs (RB, DET) — $8,800 | 22.5 proj | floor 4.9 / ceiling 42.5 | 2.56 per $1K
• Amon-Ra St. Brown (WR, DET) — $7,900 | 22.4 proj | floor 6.4 / ceiling 41.1 | 2.84 per $1K
• Derrick Henry (RB, BAL) — $7,700 | 22.3 proj | floor 7.4 / ceiling 39.0 | 2.90 per $1K
• Brock Purdy (QB, SF) — $6,500 | 22.1 proj | floor 5.6 / ceiling 39.8 | 3.40 per $1K
• Jared Goff (QB, DET) — $6,300 | 20.7 proj | floor 9.2 / ceiling 33.1 | 3.29 per $1K
• Christian McCaffrey (RB, SF) — $8,300 | 20.6 proj | floor 5.9 / ceiling 37.3 | 2.48 per $1K

Allen leads the board at 26.5 projected points in the week's second-highest market total environment (LAC @ BUF, 50.5). His ceiling of 44.8 is the widest on the list — that band width reflects his rushing floor that the Monte Carlo is picking up — and at $8,000 he's the most expensive quarterback here. If you're building around the LAC @ BUF game environment, Allen is the natural anchor.

The more interesting salary story is at quarterback. Purdy at $6,500 projects within a point of Allen (22.1 vs. 26.5) in the ARI @ SF game, where the market has SF implied at 28.5 — the highest team total on the Sunday board alongside Kansas City. Goff at $6,300 is even cheaper with a 20.7 projection and, notably, one of the tightest floor/ceiling bands in this group (9.2–33.1), which means less variance if you'd rather have a floor than a moonshot. Both open salary space for premium skill around them.

Smith-Njigba at $8,600 is the slate's priciest receiver and the only one projected above 24 points. His game (SEA @ WSH) carries a 23.5 implied total for Seattle — not a barn-burner environment, which makes the projection's source worth understanding: it's volume-driven, not game-total-driven. Henry and Gibbs are the two highest-projected backs, and both carry ceilings above 39. Henry's floor (7.4) is more comforting than Gibbs's (4.9), though Gibbs's ceiling (42.5) edges ahead — that floor gap is the risk you're accepting to reach into the DET stack.

Value per $1K — meaning projected DraftKings points divided by thousands of salary — is what tells you which names give your lineup room to spend up elsewhere. The eight names here lead the slate on that metric:

• A.J. Brown (WR, NE) — $3,000 | 13.4 proj | floor 0 / ceiling 34.1 | 4.47 per $1K
• Tank Dell (WR, HOU) — $3,000 | 11.4 proj | floor 0.3 / ceiling 23.9 | 3.80 per $1K
• Jameis Winston (QB, NYG) — $4,000 | 14.8 proj | floor 0.5 / ceiling 37.9 | 3.70 per $1K
• Brandon Aiyuk (WR, SF) — $3,000 | 10.7 proj | floor 1.3 / ceiling 22.7 | 3.57 per $1K
• Jake Tonges (TE, SF) — $2,500 | 8.9 proj | floor 0 / ceiling 33.2 | 3.56 per $1K
• Bryce Young (QB, CAR) — $5,600 | 19.9 proj | floor 7.8 / ceiling 32.7 | 3.55 per $1K
• Zach Charbonnet (RB, SEA) — $4,000 | 13.8 proj | floor 3.5 / ceiling 25.5 | 3.45 per $1K
• Tyler Shough (QB, NO) — $5,400 | 18.5 proj | floor 5.6 / ceiling 32.3 | 3.43 per $1K

A.J. Brown leads the whole value tier at 4.47 per $1K on a $3,000 salary — the floor is zero, which the band is telling you honestly, and the ceiling is 34.1. That floor reflects real volatility at a minimum-priced receiver, not a modeling artifact. If he produces anywhere near his median, the salary relief is enormous. Tonges at $2,500 is the cheapest name on the list with a ceiling of 33.2 — again, the floor is zero, and that ceiling is a low-probability tail, not an expectation. These are the names that free up salary; they are not certainties.

Bryce Young at $5,600 projects 19.9 points with a floor of 7.8 — the most comforting floor in the value tier — against a CAR @ CLE game where the market has Carolina implied at 22 points, meaning Carolina is actually the slight market favorite on scoring despite being the road team. Young is one of the few value names here with a real floor attached. Charbonnet at $4,000 projects 13.8 in a SEA offense where Smith-Njigba is getting most of the attention and salary — the two can coexist in a lineup at very different price points. Shough and Winston are the minimum-spend quarterback options if you want all your salary in skill: Shough at $5,400 with an 18.5 projection and Winston at $4,000 with 14.8 (and a 37.9 ceiling worth noting for GPP purposes).

One thing worth saying plainly: 432 draftable players carry no projection at all in our model — no NFL game history, so no number. A cheap name without a projection attached is a guess, not a value play. The value tier above is specifically the names where the model has something real to say.

Game Environments

The market has posted totals on all 16 of this week's games. The top three by total, with each team's implied total (arithmetic on the market's own posted total and spread — this is the market's math, not our projection):

1. BAL @ DAL (Sunday, 4:25 PM ET) — market total 52.5 | DAL implied 24.8, BAL implied 27.8. The week's highest-total game by a margin, and a close spread (-3 DAL). Both teams are priced to score.
2. LAC @ BUF (Sunday, 1:00 PM ET) — market total 50.5 | BUF implied 28.8, LAC implied 21.8. The market prices BUF as the week's highest-scoring individual team.
3. NYJ @ DET (Sunday, 1:00 PM ET) — market total 48.5 | DET implied 27.5, NYJ implied 21.0. Detroit is a 6.5-point home favorite with a 27.5 implied team total.

Team Stacks

The model's top five stacks by projected combined DraftKings points, with their four-player groups, combined projections, ceilings, and combined salaries:

• DET — Jahmyr Gibbs (RB), Amon-Ra St. Brown (WR), Jared Goff (QB), Sam LaPorta (TE) | proj 77.9 | ceiling 140.2 | salary $27,300 | opp implied 23.1
• BAL — Derrick Henry (RB), Zay Flowers (WR), Lamar Jackson (QB), Tyler Huntley (QB) | proj 70.2 | ceiling 142.1 | salary $26,100 | opp implied 21.4
• SF — Brock Purdy (QB), Christian McCaffrey (RB), George Kittle (TE), Mike Evans (WR) | proj 68.8 | ceiling 132.3 | salary $26,400 | opp implied 33.7
• BUF — Josh Allen (QB), James Cook III (RB), Dalton Kincaid (TE), DJ Moore (WR) | proj 65.0 | ceiling 125.9 | salary $26,200 | opp implied 37.1
• CIN — Joe Burrow (QB), Ja'Marr Chase (WR), Chase Brown (RB), Tee Higgins (WR) | proj 64.2 | ceiling 120.2 | salary $27,400 | opp implied 37.0

Detroit leads on projected points (77.9) and lives in the third-best game environment (NYJ @ DET, 48.5 total, DET implied 27.5). At $27,300 combined salary, it's the priciest of the top stacks — but the ceiling of 140.2 is essentially tied with Baltimore for first. Goff at $6,300 is doing real budget work inside that stack, which is how four players in the high-20s implied environment fit without breaking the lineup.

Baltimore's stack is unusual: it includes two quarterbacks (Lamar Jackson and Tyler Huntley), which is a GPP structure built around Jackson's dual-threat ceiling and a secondary ball-carrier option rather than a traditional skill-position stack. At $26,100 it's the cheapest of the five, and its ceiling of 142.1 is the highest on the board. The game environment — BAL @ DAL at 52.5 total, Baltimore implied at 27.8 — is the best on the slate, which is why that ceiling matters. Whether the Huntley slot makes sense depends on your lineup construction; most classic builds would swap him for a Dallas piece to catch the other side of the game total.

San Francisco's stack has the most interesting opponent context: the opp implied figure of 33.7 is the highest among all five stacks, meaning Arizona is priced to score at a high clip too. That keeps the SF side from being a pure run-it-up stack in a blowout — the market sees this as a game, not a coronation, and the 48.5 total (ARI @ SF) supports that. Purdy at $6,500 and McCaffrey at $8,300 anchor a stack that totals $26,400, with Kittle and Evans filling the remaining spots.

The BUF and CIN stacks are notable for their opponent context: BUF has an opp implied of 37.1 (LAC) and CIN an opp implied of 37.0 (PIT), meaning the model is flagging those opponents as capable of scoring enough to keep their offenses working in positive game script. The market's total for LAC @ BUF (50.5) is the second-highest of the week. The CIN stack at $27,400 is the most expensive of the five.

Two angles stand out as the week's clearest DFS cases, and here's why each one earns the attention.

First: the Detroit stack. Gibbs, St. Brown, Goff, and LaPorta project a combined 77.9 DraftKings points in a game environment where the market has Detroit implied at 27.5 against the Jets. The ceiling of 140.2 points is the kind of number that wins GPP contests. Goff at $6,300 is the salary relief that makes the stack functional — you're getting 20.7 projected points for a price you'd usually pay for a backup quarterback. The floor on Gibbs (4.9) is the honest risk to hold: this stack has real bust potential on the RB side if Detroit's game script goes sideways, and the Monte Carlo is saying so plainly. Know that going in.

Second: Brock Purdy at $6,500. He projects 22.1 points — effectively tied with the highest-salary quarterbacks on the slate — in the week's third-best game environment by team total (SF implied 28.5 in ARI @ SF). The value per $1K of 3.40 leads all quarterbacks with a meaningful projection. Pairing him with McCaffrey, Kittle, or Evans in classic build keeps your cap available for premium pieces elsewhere. His floor of 5.6 is the caution: wide band, not a safe floor, and the model is honest about the variance.

These are the projections we trust most this week, laid out with the bands they come with. No certainties here — the floor/ceiling ranges are real, and lineup construction is yours to build. The model's job is to narrow the field honestly; the rest is your call.