Tips for the week ahead

Back Armenia -0.75 on the Asian handicap at 2.56 with BetVictor — Armenia must win by more than 0.75 goals.

Our model puts the chance at 19% and the bookmakers' own consensus, margin removed, at 60%; we judge it at 44%, weighting the market more than the model, which is a fair price of 2.30. The offered 2.56 implies only 39%, so the price is 11.5% better than our estimate says it should be. That gap is the edge.

Stake 1.85 units of a 100-unit bank: the size follows the edge (a quarter-Kelly fraction), never more than 2 units.

Projected goals: Armenia 0.95, Cyprus 1.34.

The edge is an estimate and carries model error; it can be wrong, and a losing run is normal.

Back over 3.5 goals at 2.30 with Betano — 4 goals or more in the match.

Our model puts the chance at 56% and the bookmakers' own consensus, margin removed, at 42%; we judge it at 47%, weighting the market more than the model, which is a fair price of 2.11. The offered 2.30 implies only 43%, so the price is 9.2% better than our estimate says it should be. That gap is the edge.

Stake 1.77 units of a 100-unit bank: the size follows the edge (a quarter-Kelly fraction), never more than 2 units.

Projected goals: England 2.31, Croatia 1.65.

The edge is an estimate and carries model error; it can be wrong, and a losing run is normal.

Back under 2.5 goals at 2.47 with Betano — 2 goals or fewer in the match.

Our model puts the chance at 52% and the bookmakers' own consensus, margin removed, at 39%; we judge it at 44%, weighting the market more than the model, which is a fair price of 2.25. The offered 2.47 implies only 40%, so the price is 9.5% better than our estimate says it should be. That gap is the edge.

Stake 1.62 units of a 100-unit bank: the size follows the edge (a quarter-Kelly fraction), never more than 2 units.

Projected goals: Czechia 0.63, Spain 1.96.

The edge is an estimate and carries model error; it can be wrong, and a losing run is normal.

Back Kazakhstan to win at 2.80 with Dafabet.

Our model puts the chance at 47% and the bookmakers' own consensus, margin removed, at 35%; we judge it at 40%, weighting the market more than the model, which is a fair price of 2.53. The offered 2.80 implies only 36%, so the price is 10.7% better than our estimate says it should be. That gap is the edge.

Stake 1.48 units of a 100-unit bank: the size follows the edge (a quarter-Kelly fraction), never more than 2 units.

Projected goals: Moldova 1.14, Kazakhstan 1.58.

The edge is an estimate and carries model error; it can be wrong, and a losing run is normal.

Back Switzerland +0.25 on the Asian handicap at 1.42 with Bet365 — Switzerland starts 0.25 goals ahead, so a narrow defeat can still win the bet.

Our model puts the chance at 80% and the bookmakers' own consensus, margin removed, at 77%; we judge it at 78%, weighting the market more than the model, which is a fair price of 1.28. The offered 1.42 implies only 70%, so the price is 11.1% better than our estimate says it should be. That gap is the edge.

Stake 2.00 units of a 100-unit bank: the size follows the edge (a quarter-Kelly fraction), never more than 2 units.

Projected goals: Slovenia 0.75, Switzerland 1.57.

The edge is an estimate and carries model error; it can be wrong, and a losing run is normal.

Back under 1.5 goals at 3.00 with Bet365 — 1 goal or fewer in the match.

Our model puts the chance at 45% and the bookmakers' own consensus, margin removed, at 32%; we judge it at 37%, weighting the market more than the model, which is a fair price of 2.68. The offered 3.00 implies only 33%, so the price is 11.8% better than our estimate says it should be. That gap is the edge.

Stake 1.48 units of a 100-unit bank: the size follows the edge (a quarter-Kelly fraction), never more than 2 units.

Projected goals: Luxembourg 0.75, Estonia 1.06.

The edge is an estimate and carries model error; it can be wrong, and a losing run is normal.