Scoreline picks that maximise contest points — not headlines. [ market 70% · form 30% · ranking 15% ]
FIFA № 30 (1559 pts) — vs — FIFA № 64 (~1390 pts)
Marsch's press meets Barbarez's bunker. 76% of Canada's matches went under 2.5 — and Bosnia have scored 0 or 1 in five straight.
Grid covers 97.6% of outcomes · over 2.5 goals: 42.6%
| Score | P(exact) | E[points] |
|---|---|---|
| 1–0◂ PLAY | 13.7% | 1.487 |
| 2–0 | 11.1% | 1.408 |
| 2–1 | 9.2% | 1.351 |
| 3–0 | 5.7% | 1.247 |
| 3–1 | 4.8% | 1.218 |
| 4–0 | 2.2% | 1.141 |
FIFA № 17 (1671 pts) — vs — FIFA № 41 (1505 pts)
1–1 is a near-tie on probability (13.1%) but the winner-points backup decides it. Every Paraguay result in 12 months came by a one-goal margin.
Grid covers 98.2% of outcomes · over 2.5 goals: 41.3%
| Score | P(exact) | E[points] |
|---|---|---|
| 1–0◂ PLAY | 13.3% | 1.380 |
| 2–0 | 10.0% | 1.281 |
| 2–1 | 9.0% | 1.251 |
| 3–0 | 4.8% | 1.125 |
| 3–1 | 4.3% | 1.111 |
| 3–2 | 1.9% | 1.040 |
Closing 1X2 and over/under 2.5 odds are the strongest free predictor in football. We strip the bookmaker margin to recover true outcome probabilities, then nudge them with twelve-month form (30%) and the FIFA ranking's Elo expectancy (15%).
A Poisson score model with the Dixon-Coles low-score correction ρ recovers each side's expected goals from those probabilities — giving a probability for every scoreline, including the draws plain Poisson gets wrong.
Every scoreline is ranked by P(exact)·5 + P(winner, not exact)·2. The best pick is often not the most likely score — see Folio II, where 1–1 nearly matches 1–0 on probability but loses on expected points.
Lock late. These numbers use mid-day odds. Line-ups drop ±70 minutes before kick-off and the market reprices — re-run the agent just before your contest locks. Scoring assumed: exact 5 pts, winner 2 pts; if your contest differs, the optimal picks may shift.