World Cup 2026 Matchday 9 Recap: Brazil Cashes, but Two Goal Bets Miss on a −2.7u Day

An honest down day on the World Cup 2026 model: matchday 9 went 1-2 for −2.7 units, our one winner (Brazil) outweighed by two losing goal bets. The tournament book slips from +6.92u to +4.50u. As ever, though, we look past the daily P&L to the metric that actually predicts profit — and the closing-line value held up: across the settled book our average CLV sits at +2.5pp, nudged *up* by two sharp prices that lost to variance. The model's directional read went 2-of-4, and the two misses tell two very different stories — one was the guardrail working, one was a clean upset. Here is all of it.

USA 2–0 Australia — the model leaned the wrong way, and the guardrail stopped us betting it

Result: USA 2–0 Australia. Pre-match model: USA win 40% / Draw 14% / Australia win 45%. Brier: 0.195. This is the fixture worth being most transparent about. Our national-team model — which leans on Elo — actually edged Australia ahead, while the sharp market made the United States overwhelming ~80% favourites. That is a huge model-vs-market gap, and our system flagged it automatically: a co-host playing at home, with a squad full of top-five-league talent, is exactly the kind of edge an Elo lens cannot see. We published no confident winner call and placed no bet. The USA then won 2–0, comfortably, as the market expected.

So the model's raw read was wrong — and it cost us nothing, because the disagreement guard did precisely its job. It is worth naming the blind spot plainly: at a World Cup our international model gives the host nations no home-field bump (every match is treated as neutral), and it cannot price the gap between a deep CONCACAF co-host and a young, rebuilding Socceroos side. We flag those games, we defer to the market on the winner, and we are actively reviewing whether the hosts deserve a structural home edge. Full numbers in the USA vs Australia match centre; our pre-match note is here.

Brazil 3–0 Haiti — the cleanest call of the day, and our one winner

Result: Brazil 3–0 Haiti. Pre-match model: Brazil win 72% / Draw 14% / Haiti win 14%. Brier: 0.038 — the sharpest read of the matchday. This was never a winner question; it was a goals question. Brazil were always going to win, so the value sat in the total: our model projected close to three goals but liked the price on the Under against a line stretched by Brazil's reputation. A 3–0 — exactly three goals — cashed the Under 3.5 comfortably. A clean, low-variance win on the day's most predictable fixture.

Scotland 0–1 Morocco — right on the winner, wrong on the goals

Result: Scotland 0–1 Morocco. Pre-match model: Scotland win 29% / Draw 16% / Morocco win 55%. Brier: 0.103. The winner read was directionally correct — Morocco were the model's pick and Morocco won — although here the model was the *under*-confident one (55% to the market's ~83%), so we placed no winner bet. Our value sat in the goals market, an Over 2.5 lean, and a tight 0–1 grind buried it: just one goal in a cagey, low-event game. A fair miss to own. The price was good — +5.5pp of closing-line value — but the result was the kind of stalemate the Over needs to avoid.

Türkiye 0–1 Paraguay — a clean miss, no guardrail to hide behind

Result: Türkiye 0–1 Paraguay. Pre-match model: Türkiye win 66% / Draw 14% / Paraguay win 20%. Brier: 0.362 — the model's worst call of the matchday, and unlike USA–Australia there was no disagreement flag here: the model and the market both favoured Türkiye. Paraguay's disciplined, defensively drilled side won it 1–0, and our Over 2.5 lean — the day's highest-EV ticket at +17.5% — went down with a single goal. This was a genuine miss on both fronts, the winner read and the goals bet, with no protection policy to soften it. Some upsets you simply eat. Türkiye, beaten in both of their group games, are out — eliminated at the group stage; details in the Türkiye vs Paraguay match centre.

Matchday 9 scorecard: 2-of-4, at the model's calibration ceiling

The full matchday 9 1X2 breakdown:

• Brazil 3–0 Haiti — model leaned Brazil (72%), result: BRAZIL. Brier 0.038. ✓
• Scotland 0–1 Morocco — model leaned Morocco (55%), result: MOROCCO. Brier 0.103. ✓
• USA 2–0 Australia — model leaned Australia (45%), result: USA. Brier 0.195. ✗ (disagreement-flagged — no bet placed)
• Türkiye 0–1 Paraguay — model leaned Türkiye (66%), result: PARAGUAY. Brier 0.362. ✗ (genuine upset)
• Average 1X2 Brier ≈ 0.175 across the four — right on the model's long-run calibrated figure (0.177), despite a four-goal swing of variance between the best and worst calls.

The throughline is the one we keep returning to: our national-team model reads results through Elo, which cannot see club-level quality, a host nation's home edge, or the strength gap between confederations. The USA miss came straight from that blind spot — and we flagged it and routed around it on the winner market. The Türkiye miss did not; it was a favourite that simply lost. Two different misses, handled two different ways, and the aggregate accuracy still landed exactly where the model says it should.

Running tournament record: 10 wins / 8 losses / 2 voids, +4.50 units

Here is the full cumulative World Cup 2026 value-bet record through matchday 9. Convention as always: voids are counted separately — a void returns the stake in full and is never a loss; the win rate is over settled wins and losses only; and units are fractional-Kelly stakes recomputed from the database (1 unit = 1% of bankroll), so they match the public track record.

• Matchday 1 — South Korea Over 2.5 @ 2.29 (+9.1% EV). WON ✓ — +2.27u.
• Matchday 2 — USA Over 2.5 @ 2.42 (+15.3% EV). WON ✓ — +3.82u. USA AH +0 @ 1.556 (+6.2% EV). WON ✓ — +1.56u.
• Matchday 3 — Switzerland Under 2.75 @ 2.03 (+5.7% EV). WON ✓ — +1.43u.
• Matchday 4 — Sweden AH +0 @ 1.493 (+9.4% EV). WON ✓ — +2.36u. Netherlands AH +0 @ 1.583. VOID — stake returned.
• Matchday 5 — Spain Under 3.5 @ 1.97 (+12.1% EV). WON ✓ — +3.02u. Belgium AH +0 @ 1.32. VOID — stake returned. Uruguay to win @ 1.50 (+8.4% EV). LOST ✗ — −4.20u. Saudi Arabia vs Uruguay Over 2.25 @ 1.85 (+5.6% EV). LOST ✗ — −0.82u (half-loss).
• Matchday 6 — Austria Under 2.5 @ 2.12 (+7.1% EV). LOST ✗ — −1.58u.
• Matchday 7 — DR Congo +1.5 @ 2.00 (+7.0% EV). WON ✓ — +1.76u. England to win @ 1.75 (+6.8% EV). WON ✓ — +1.70u.
• Matchday 8 — Switzerland AH +0 @ 1.271 (+5.8% EV). WON ✓ — +1.35u. Canada Under 2.5 @ 2.16 (+11.5% EV). LOST ✗ — −2.47u. Mexico vs South Korea Over 2.25 @ 2.09 (+8.0% EV). LOST ✗ — −1.83u. South Korea +0.5 @ 1.87 (+5.1% EV). LOST ✗ — −1.45u.
• Matchday 9 — Brazil Under 3.5 @ 2.10 (+9.2% EV). WON ✓ — +2.30u. Scotland vs Morocco Over 2.5 @ 2.30 (+8.5% EV). LOST ✗ — −1.63u. Türkiye vs Paraguay Over 2.5 @ 2.30 (+17.5% EV). LOST ✗ — −3.36u. (USA vs Australia — no bet; model-market disagreement, deferred to the market.)
• Cumulative: 10 wins / 8 losses / 2 voids (55.6% win rate on settled bets). +4.50 units. Average closing-line value across the settled book +2.5pp.

Two down days in a row have trimmed the book from its +11u peak, and we are not dressing that up. But the texture matters: matchday 9's two losers carried +5.5pp and a strong-EV price, and the book's average CLV ticked up, not down. Over a couple of hundred bets a stretch like this is variance; over a full tournament the CLV is the signal, and it remains firmly positive. These are settled historical results published as a transparent record — not live tips. Tonight's actionable value bets are members-only.

What's next: tonight's previews

Matchday 10 rolls through Groups E and F. Our full pre-match model breakdowns are live now:

• Netherlands vs Sweden prediction — a Group F clash with the Dutch favoured but far from clear of a physical Sweden, and a match where the model has flagged value for members.
• Germany vs Ivory Coast prediction — Germany strong favourites in Group E with the day's biggest expected-goals number.
• Ecuador vs Curaçao prediction — Ecuador heavy favourites against the smallest nation ever to reach a World Cup.
• Tunisia vs Japan prediction — Japan a clear Group F favourite away from home, and the other match where the model sees value for members.

Follow all group standings, model probabilities and the live bracket on the World Cup 2026 hub. Yesterday's matchday 8 recap is also live: World Cup 2026 Matchday 8 Recap.