# Week 1 (Sep 3-7, 2026) — Steele × Connelly × Market Deep Dive

**Generated:** 2026-08-15 · **Basis:** preseason SP+ (live CFBD) · **Threshold:** 3.0
**Sources:** Phil Steele 2026 Digital Magazine (team pages), Connelly 2026 ESPN+ previews
(projections/qb_situations/continuity), CFBD lines movement (Bovada + DraftKings open vs current),
picks_latest.json (2026-08-15T06:27:22Z run, 16 picks).

---

## 1. The model card (16 picks, all week 1)

| # | Game | Proj | Line | Edge | Pick | Bucket | Bucket acc |
|---|---|---|---|---|---|---|---|
| 1 | Northern Illinois @ Iowa | +34.3 | -30.25 | +4.05 | Iowa -30.25 | 3-7 | 0.5118 |
| 2 | Texas State @ Texas | +27.4 | -30.5 | -3.10 | Texas State +30.5 | 3-7 | 0.5118 |
| 3 | East Carolina @ Alabama | +22.4 | -27.0 | -4.60 | East Carolina +27.0 | 3-7 | 0.5118 |
| 4 | Tulane @ Duke | +12.4 | -8.5 | +3.90 | Duke -8.5 | 3-7 | 0.5118 |
| 5 | Oregon State @ Houston | +15.5 | -19.25 | -3.75 | Oregon State +19.25 | 3-7 | 0.5118 |
| 6 | Coastal Carolina @ West Virginia | +15.9 | -19.5 | -3.60 | Coastal Carolina +19.5 | 3-7 | 0.5118 |
| 7 | Arkansas State @ Memphis | +14.4 | -10.0 | +4.40 | Memphis -10.0 | 3-7 | 0.5118 |
| 8 | UL Monroe @ Mississippi State | +36.2 | -28.5 | +7.70 | Mississippi State -28.5 | 7-14 | 0.4356 |
| 9 | Massachusetts @ Rutgers | +38.1 | -30.5 | +7.60 | Rutgers -30.5 | 7-14 | 0.4356 |
| 10 | Missouri State @ Texas A&M | +42.5 | -39.0 | +3.50 | Texas A&M -39.0 | 3-7 | 0.5118 |
| 11 | Sam Houston @ Troy | +20.5 | -16.25 | +4.25 | Troy -16.25 | 3-7 | 0.5118 |
| 12 | Miami (OH) @ Pittsburgh | +12.7 | -16.25 | -3.55 | Miami (OH) +16.25 | 3-7 | 0.5118 |
| 13 | Akron @ Wake Forest | +28.8 | -22.5 | +6.30 | Wake Forest -22.5 | 3-7 | 0.5118 |
| 14 | Ball State @ Ohio State | +60.3 | -50.25 | +10.05 | Ohio State -50.25 | 7-14 | 0.4356 |
| 15 | Central Michigan @ New Mexico | +17.6 | -12.75 | +4.85 | New Mexico -12.75 | 3-7 | 0.5118 |
| 16 | Toledo @ Michigan State | +13.8 | -10.5 | +3.30 | Michigan State -10.5 | 3-7 | 0.5118 |

**Bucket honesty (from the engine's own backtest):** 3-7 bucket hits 51.18% — below the
52.38% break-even at -110. 7-14 bucket hits 43.56% — **loses money**. The only historically
validated signal (edge ≥ 12, 55.8% cover, no losing season in 8 years) has ZERO qualifiers
this week (max edge: Ohio State +10.05). This is a small-stakes-or-pass card.

---

## 2. Steele × Connelly intel per pick

### Tier 1 — model + qualitative + market all agree (5 plays)

**New Mexico -12.75 vs Central Michigan (+4.85)**
- Steele: "return QB Layne... now veteran O-line and an upgraded receiving corps. I look for improved numbers in 2026." Defense returns MW Def POY Jaxton Eck + 6 of top 10 tacklers.
- Connelly: SP+ -0.5, 8.0 wins, 63% returning production (#8 FBS), QB "won last 6 regular season games, 4 starting linemen return including all-conf frosh Robnett."
- CMU (Steele p324): loses QB Labas, top 2 rushers, "most likely #2 WR"; defense loses 8 starters and 9 of top 10 tacklers incl. top 6. Switching to Power Spread — year 1 of a rebuild.
- Market: steamed TOWARD UNM (Bovada -12.5 open → -13.0 current).

**Wake Forest -22.5 vs Akron (+6.30)**
- Steele: "This year including transfers 9 starters are back so the defense will be the strong suit in 2026. DC Hazelton did a great job... now has more talent to work with." QB Gio Lopez "thrived under OC Ezell at South Alabama in '24... gives the team the pass threat they lacked."
- Connelly: SP+ 3.5, 6.2 wins, 53% ret. Akron: SP+ -19.4, 4.2 wins, 40% ret, QB Poffenbarger on 5th school in 6 years.
- Market: flat -22.5 both books.

**Troy -16.25 vs Sam Houston (+4.25)**
- Steele: new "Veer and Shoot" offense from Adam Austin (Tarleton St); "return two long time starters and should be improved."
- Connelly: Troy SP+ -6.1, 7.3 wins, both QBs back (20 combined starts, 3,962 pass yds). SHSU SP+ -26.2, 2.8 wins, 49% ret, "counting on sophomore returnees... 2nd straight transition year."
- Market: slightly AGAINST Troy (DK -17.5 open → -16.5 current; Bovada -17.0 → -16.0).

**Texas A&M -39.0 vs Missouri State (+3.50)**
- Steele (MSU p358): "This year they have just 2 starters back and lose a 1,000 yard RB, QB Clark and all 5 OL and figure to take another step back offensively... DL and CB rooms in total rebuild."
- Connelly: MSU SP+ -18.7, 3.9 wins, **34% ret (#134-ish, lowest on card)**, "almost entirely new offense." A&M SP+ 20.3, 8.4 wins, 65% ret; QB "3000+ pass yds 550+ rush" but 12 INT (3rd most SEC) — the one crack.
- Market: steamed TOWARD A&M (Bovada -38.5 → -39.5).

**Iowa -30.25 vs Northern Illinois (+4.05)**
- Steele: "They lose QB Gronowski and #2, #3 and #5 DC OL's. There is plenty of competition and they will not drop off much with my computer calling for 28.9 ppg." Parker D "will have them among the best as always."
- Connelly: Iowa SP+ 13.6, 8.0 wins, 44% ret; "basic competence needed" at QB after Gronowski. NIU: SP+ -18.2 (CFBD), no Connelly combined row, Steele p280.
- Market: steamed TOWARD Iowa (Bovada -29.5 → -31.0).

### Tier 2 — model + one layer, with a flag (6 plays)

**Texas State +30.5 @ Texas (-3.10)**
- Connelly: QB Jackson "4th-best QBR of any G6 QB; best of any returnee; 2 favorite targets return." Steele: "QB Jackson and his top 4 receivers are all back... they should be even more potent in 2026" — 36.5 ppg last year. Best cover case on the card.
- Texas: 23.5 SP+, 8.6 wins, 68% ret — elite, and Steele notes QB "weakness is accuracy throwing on the run."
- Market: flat -30.5.

**East Carolina +27.0 @ Alabama (-4.60)**
- Connelly: ECU 7.8 wins projection, QB battle (Griffis scrambler-gambler vs Williams pocket). Steele: "They did lose their QB to Illinois and only have 1 starter back from the '25 offense. This figure to take a step back but will still be potent."
- Alabama: 18.0 SP+, QB Russell "nearly perfect in tiny sample; massive upside."
- Market: steamed AGAINST the pick — Bovada Bama -25.5 → -28.5. Steele's "step back" + market steam both undercut the cover.

**Duke -8.5 vs Tulane (+3.90)**
- Steele: "The unexpectedly lost Mensah and top WR Barkate at the end of the portal to Miami... They will still take a step back in offensive production (computer calls for 29.3 ppg)."
- Connelly: Duke QB "threw for 3000 yds with torn ACL last season; replaces Mensah" (Eget). Tulane SP+ -5.5, 6.5 wins, QB Chriss-Gremillion "electric runner (22 mph), 13 career starts"; Steele: "they will not fall off much."
- Market: **moved AGAINST Duke** — Bovada -10.0 → -8.5.

**Oregon State +19.25 @ Houston (-3.75)**
- Steele (Houston): "They have QB Weigman, their top WR Thomas a top notch O-line and add in two time 1300+ rusher Makhi Hughes who played for Fritz at Tulane so are even stronger and will improve again." Defense: 6 starters back, "even tougher."
- OSU (Steele): new offense, "The offense can only improve" — but "few known difference-makers"; DC MacIntyre hire noted.
- Connelly: Houston 8.2 SP+, 7.4 wins, 65% ret. OSU -6.4, 5.1 wins, 47% ret.
- Market: steamed AGAINST the pick — Bovada Houston -18.5 → -20.0.

**Coastal Carolina +19.5 @ West Virginia (-3.60)**
- Steele (WVU): "This year 17 of the top 18 tacklers are gone" — the best structural fade case on the card. But Rich Rod offense "could top that by a TD" after 21.8 ppg; O-line gets Rick Trickett.
- Connelly: CCU -13.8 SP+, 5.2 wins, QB battle (Bailey followed Beard from Missouri St). WVU 0.6 SP+, 5.2 wins, 52% ret.
- Market: steamed AGAINST the pick — Bovada WVU -17.5 → -21.5 (sharpest move of the slate).

**Miami (OH) +16.25 @ Pittsburgh (-3.55)**
- Connelly: Miami 8.4 wins, 49% ret; QB room "no clear spring winner" (Gotkowski/McComb/Heavner). Steele (Pitt): OC Bell year 3 — "the last time they had an OC for 3 years they won the ACC title"; QB "57.6 QBR as true freshman"; Narduzzi D "always crafts a solid defense."
- Market: flat -16.5. Model's 12.7 projection is the only pro-Miami voice; Steele's Pitt bull case is real.

### Tier 3 — big edges, losing bucket (3 plays — pass or tiny)

**Mississippi State -28.5 vs UL Monroe (+7.70)**
- Steele: "the Kamario Taylor era begins... if Taylor stays healthy he should have this offense improve again." ULM (Connelly): -24.1 SP+, 2.6 wins, "RB corps completely wiped out again."
- Bucket 7-14 hits 43.56%. Big number on a first-year starting QB. Pass.

**Rutgers -30.5 vs Massachusetts (+7.60)**
- Steele: "with a softer schedule, they should be improved... new QB has 1,341-yard RB Raymond, top WR Duff and a solid offensive line." UMass: -30.9 SP+, 2.9 wins, "easiest schedule in FBS," QB Watson (VT reserve) vs Johnson (FAMU).
- Everything agrees — which is the trap. 30.5 points is a lot for a team that scored 29 ppg last year. Bucket 7-14. Pass.

**Ohio State -50.25 vs Ball State (+10.05)**
- Steele: "they have QB Sayin, RB Jackson and WR Smith, plus a veteran Oline with 4 returning starters and should take the reins off Sayin and be one of the nations best." BUT: "This year they lose 7 NFL DC's, including 3 first rounders (all in the Top 11!). The talent is there but it will be tough to match last year's spectacular numbers vs a tougher schedule."
- Ball State: "much stronger on offense" (year 2), defense "lacks FBS experience" but "kept them in most games." Steele projects improvement on both sides.
- Biggest edge on the card, worst bucket on the card, against a team that got better. Pass.

### No-bucket note (Michigan State -10.5, +3.30)
- Steele: "just two starters from that unit is back" on OL; D loses 13 of top 16 tacklers; "will need the Spartans to have hit some home runs in the portal to improve much." Toledo: 33% ret, "less experienced while also learning a new offense... will be down a notch" (Steele).
- Weakest edge of the 3-7 group; both teams rebuilding. Leans pass at $20.

---

## 3. The read

**Five flat plays, $20-25 stakes:** New Mexico -12.75, Wake Forest -22.5, Troy -16.25,
Texas A&M -39.0, Iowa -30.25. Model edge + qualitative layer + (mostly) market agreement.

**Skip:** the three 7-14 bucket plays (MSU/Rutgers/OSU) — the engine's own backtest says
that bucket loses money, and Steele/Connelly agreeing with the favorites is exactly the
consensus trap. Skip Michigan State too (weakest 3-7 edge, both sides rebuilding).

**Respect market pushback:** 4 of 16 picks have market movement AGAINST the model (Duke,
Memphis, Oregon State, Coastal Carolina). The market is pricing the same Steele/Connelly
signals the model doesn't see. Don't fight it at $20 stakes.

**Structural cautions:** OSU -50.25 against a team that improved (and OSU replacing 3
first-round DL); Rutgers -30.5 against a defense that allowed 31.8 ppg vs a tough schedule
(softer schedule ≠ blowout); A&M -39.0 against the QB's 12-INT, 15th-of-16 catchable rate.

---

## 4. Caveats

- Northern Illinois and Central Michigan are absent from connelly combined CSVs (checked
  all_projections/qb_situations/continuity) — NIU/CMU rows exist in per-conference MAC
  files; SP+ values from CFBD (NIU -18.2, CMU -12.4) used instead.
- Steele forecast blocks extracted for 21 of 32 teams; pages for UMass, Miami (OH), Akron,
  Texas, Coastal Carolina, Arkansas State, Alabama, UL Monroe, NIU, Sam Houston, Texas A&M
  had no "PHIL'S FORECAST" block in extraction (QB/position outlook text used where present).
- Lines move. Verify every number at bet time; the movement column above is open vs
  current as of 2026-08-15.
- Model output is not betting advice. Bucket hit rates are the engine's own backtest
  (2025); several buckets lose money at -110. Flat units only.