The same production stack priced under 4 course readings, against the bookmaker's pre-stage line. Each scenario is one opinion about what kind of day this is; the results section settles the argument.
The fork
Course profile is an input the model can't see from a map — it's set from reference races (or overridden by hand). Different readings redistribute the same probability mass across rider types.
The board
Blend win probability per scenario (cascade × LambdaRank, DNF-adjusted at the pricing layer), next to the devigged Kambi (Unibet) win line captured the morning before the stage. "Max edge" compares the most favourable scenario against the book — treat it as scenario-conditional, not a free bet.
Read "Max edge" as directional, not as a price. The blend is a set of per-rider marginals and does not sum to one book — across this field each reading totals Versatile sprint 98%, Pauls sticks 100%, Full sprint 101%, Hindsight (post-hoc) 104% — while the devigged book column sums to exactly 100%. Putting the two side by side flatters the model, and Max edge is inflated by about that factor: divide a rider's probability by his reading's total before calling anything an edge. Of our heads only the LambdaRank one is natively book-preserving.
| Rider | Team | Versatile sprint | Pauls sticks | Full sprint | Hindsight (post-hoc) | Book odds | Book devig | Max edge |
|---|---|---|---|---|---|---|---|---|
| Brennan Matthew | Team Visma | Lease a Bike | 45.6% | 26.2% | 43.0% | 46.0% | 1.85 | 35.9% | +10.2pp |
| Meeus Jordi | Red Bull-BORA-hansgrohe | 10.0% | 2.9% | 16.6% | 11.7% | 13.00 | 5.1% | +11.5pp |
| Pogacar Tadej | UAE Team Emirates-XRG | 1.8% | 11.1% | 1.0% | 2.4% | 21.00 | 3.2% | +7.9pp |
| Laporte Christophe | Team Visma | Lease a Bike | 5.1% | 9.2% | 4.6% | 6.0% | 41.00 | 1.6% | +7.6pp |
| van Aert Wout | Team Visma | Lease a Bike | 8.4% | 6.1% | 6.1% | 8.0% | 6.00 | 11.1% | -2.7pp |
| Pedersen Mads | Lidl-Trek | 4.9% | 8.3% | 4.9% | 5.8% | 9.00 | 7.4% | +1.0pp |
| Turner Ben | Netcompany INEOS | 3.6% | 4.7% | 2.5% | 4.1% | 41.00 | 1.6% | +3.1pp |
| Laurance Axel | Netcompany INEOS | 2.1% | 4.1% | 1.2% | 1.9% | 41.00 | 1.6% | +2.5pp |
| De Schuyteneer Steffen | Lotto-Intermarché | 2.3% | 1.2% | 4.1% | 2.5% | 101.00 | 0.7% | +3.4pp |
| Dainese Alberto | Soudal Quick-Step | 1.4% | 0.2% | 3.0% | 1.6% | 51.00 | 1.3% | +1.7pp |
| Fisher-Black Finn | Red Bull-BORA-hansgrohe | 0.2% | 2.8% | 0.0% | 0.2% | 67.00 | 1.0% | +1.8pp |
| Groves Kaden | Alpecin-Premier Tech | 1.1% | 0.3% | 2.1% | 1.3% | 19.00 | 3.5% | -1.3pp |
| Aular Orluis | Movistar Team | 1.2% | 1.6% | 1.1% | 1.2% | 51.00 | 1.3% | +0.3pp |
| Canal Carlos | Movistar Team | 0.4% | 1.4% | 0.3% | 0.4% | 301.00 | 0.2% | +1.2pp |
| Renard Alexis | Cofidis | 1.1% | 1.3% | 0.6% | 1.1% | 126.00 | 0.5% | +0.8pp |
| Coquard Bryan | Cofidis | 0.9% | 0.2% | 1.3% | 1.1% | 51.00 | 1.3% | +0.0pp |
| Brustenga Marc | Equipo Kern Pharma | 0.3% | 0.1% | 1.1% | 0.3% | 251.00 | 0.3% | +0.8pp |
| Govekar Matevz | Bahrain Victorious | 1.0% | 0.2% | 1.0% | 1.0% | 201.00 | 0.3% | +0.7pp |
| # | Rider | Blend | Cascade | LambdaRank | Podium | Top 15 | Book odds |
|---|---|---|---|---|---|---|---|
| 1 | Brennan Matthew | 45.6% | 40.1% | 47.7% | 59.8% | 80.3% | 1.85 |
| 2 | Meeus Jordi | 10.0% | 11.2% | 9.9% | 33.1% | 67.0% | 13.00 |
| 3 | van Aert Wout | 8.4% | 7.4% | 8.8% | 26.4% | 57.0% | 6.00 |
| 4 | Laporte Christophe | 5.1% | 7.2% | 4.7% | 26.3% | 64.4% | 41.00 |
| 5 | Pedersen Mads | 4.9% | 6.3% | 4.7% | 21.8% | 60.0% | 9.00 |
| 6 | Turner Ben | 3.6% | 3.4% | 3.7% | 14.0% | 53.4% | 41.00 |
| 7 | De Schuyteneer Steffen | 2.3% | 2.3% | 2.3% | 11.1% | 51.8% | 101.00 |
| 8 | Laurance Axel | 2.1% | 2.5% | 2.1% | 8.7% | 48.7% | 41.00 |
| 9 | Pogacar Tadej | 1.8% | 3.3% | 1.5% | 9.9% | 28.7% | 21.00 |
| 10 | Dainese Alberto | 1.4% | 1.0% | 1.5% | 5.2% | 32.4% | 51.00 |
| 11 | Aular Orluis | 1.2% | 1.4% | 1.2% | 5.9% | 40.7% | 51.00 |
| 12 | Groves Kaden | 1.1% | 1.2% | 1.1% | 4.8% | 29.5% | 19.00 |
| 13 | Renard Alexis | 1.1% | 1.5% | 0.9% | 7.2% | 42.6% | 126.00 |
| 14 | Govekar Matevz | 1.0% | 1.9% | 0.7% | 8.3% | 43.3% | 201.00 |
| 15 | Hofstetter Hugo | 0.9% | 1.2% | 0.8% | 6.9% | 51.7% | 41.00 |
| 16 | Coquard Bryan | 0.9% | 0.4% | 1.0% | 1.6% | 59.8% | 51.00 |
| 17 | Miquel Delgado Pau | 0.7% | 0.7% | 0.7% | 4.2% | 42.3% | 34.00 |
| 18 | González David | 0.5% | 0.1% | 0.6% | 1.3% | 47.1% | 151.00 |
| 19 | Canal Carlos | 0.4% | 0.3% | 0.4% | 2.5% | 29.7% | 301.00 |
| 20 | Brustenga Marc | 0.3% | 0.1% | 0.3% | 2.3% | 34.0% | 251.00 |
| 21 | Cort Magnus | 0.3% | 0.1% | 0.3% | 0.7% | 9.3% | 67.00 |
| 22 | Braet Vito | 0.2% | 0.1% | 0.3% | 1.0% | 42.9% | 81.00 |
| 23 | Macias Cesar | 0.2% | 0.2% | 0.2% | 2.5% | 35.1% | 251.00 |
| 24 | Mulubrhan Henok | 0.2% | 0.1% | 0.2% | 0.9% | 9.0% | 201.00 |
| 25 | Oliveira Ivo | 0.2% | 0.2% | 0.2% | 1.5% | 11.2% | 201.00 |
| 26 | Albanese Vincenzo | 0.2% | 0.2% | 0.1% | 0.9% | 20.5% | 201.00 |
| 27 | Garofoli Gianmarco | 0.2% | 0.2% | 0.2% | 0.9% | 12.3% | 401.00 |
| 28 | Fisher-Black Finn | 0.2% | 0.1% | 0.2% | 2.0% | 9.4% | 67.00 |
| 29 | Nys Thibau | 0.1% | 0.2% | 0.1% | 0.6% | 7.2% | 101.00 |
| 30 | Tulett Ben | 0.1% | 0.2% | 0.1% | 1.1% | 9.5% | 201.00 |
| # | Rider | Blend | Cascade | LambdaRank | Podium | Top 15 | Book odds |
|---|---|---|---|---|---|---|---|
| 1 | Brennan Matthew | 26.2% | 26.5% | 26.0% | 42.1% | 57.4% | 1.85 |
| 2 | Pogacar Tadej | 11.1% | 14.9% | 10.1% | 34.9% | 56.9% | 21.00 |
| 3 | Laporte Christophe | 9.2% | 10.6% | 8.8% | 27.9% | 53.6% | 41.00 |
| 4 | Pedersen Mads | 8.3% | 9.9% | 7.9% | 26.6% | 55.9% | 9.00 |
| 5 | van Aert Wout | 6.1% | 6.0% | 6.1% | 20.2% | 42.6% | 6.00 |
| 6 | Turner Ben | 4.7% | 3.8% | 4.9% | 14.9% | 40.1% | 41.00 |
| 7 | Laurance Axel | 4.1% | 3.6% | 4.3% | 13.4% | 47.0% | 41.00 |
| 8 | Meeus Jordi | 2.9% | 2.4% | 3.0% | 11.9% | 26.9% | 13.00 |
| 9 | Fisher-Black Finn | 2.8% | 5.3% | 2.2% | 15.1% | 38.7% | 67.00 |
| 10 | Aular Orluis | 1.6% | 2.1% | 1.4% | 6.7% | 33.8% | 51.00 |
| 11 | Canal Carlos | 1.4% | 1.2% | 1.5% | 4.4% | 28.4% | 301.00 |
| 12 | Renard Alexis | 1.3% | 1.3% | 1.3% | 5.3% | 23.7% | 126.00 |
| 13 | De Schuyteneer Steffen | 1.2% | 1.4% | 1.1% | 5.0% | 20.1% | 101.00 |
| 14 | Tulett Ben | 0.9% | 1.6% | 0.7% | 5.0% | 27.2% | 201.00 |
| 15 | Onley Oscar | 0.9% | 0.2% | 1.1% | 4.5% | 23.7% | 251.00 |
| 16 | Miquel Delgado Pau | 0.8% | 0.9% | 0.8% | 4.6% | 28.1% | 34.00 |
| 17 | Carapaz Richard | 0.6% | 0.1% | 0.7% | 1.1% | 17.1% | 301.00 |
| 18 | Johannessen Tobias Halland | 0.6% | 0.1% | 0.7% | 2.7% | 17.2% | 126.00 |
| 19 | Mulubrhan Henok | 0.6% | 0.2% | 0.6% | 1.2% | 22.5% | 201.00 |
| 20 | Debruyne Ramses | 0.5% | 0.3% | 0.6% | 2.2% | 22.6% | 301.00 |
| 21 | Cobo Ivan | 0.5% | 0.2% | 0.6% | 2.3% | 9.8% | 501.00 |
| 22 | Skjelmose Mattias | 0.5% | 0.8% | 0.5% | 2.8% | 20.2% | 501.00 |
| 23 | Roglic Primoz | 0.5% | 0.3% | 0.6% | 1.8% | 14.3% | 501.00 |
| 24 | Brenner Marco | 0.5% | 0.2% | 0.6% | 1.4% | 25.0% | 401.00 |
| 25 | Berthet Clément | 0.5% | 0.1% | 0.6% | 1.1% | 16.3% | n/a |
| 26 | Hofstetter Hugo | 0.4% | 0.8% | 0.3% | 3.5% | 22.7% | 41.00 |
| 27 | Garofoli Gianmarco | 0.4% | 0.1% | 0.4% | 1.4% | 20.1% | 401.00 |
| 28 | Kron Andreas | 0.4% | 0.0% | 0.4% | 0.8% | 9.9% | 101.00 |
| 29 | Tejada Harold | 0.3% | 0.2% | 0.4% | 0.7% | 11.3% | 501.00 |
| 30 | Groves Kaden | 0.3% | 0.4% | 0.3% | 1.1% | 4.5% | 19.00 |
| # | Rider | Blend | Cascade | LambdaRank | Podium | Top 15 | Book odds |
|---|---|---|---|---|---|---|---|
| 1 | Brennan Matthew | 43.0% | 42.1% | 42.7% | 63.9% | 87.4% | 1.85 |
| 2 | Meeus Jordi | 16.6% | 13.9% | 17.1% | 43.7% | 84.0% | 13.00 |
| 3 | van Aert Wout | 6.1% | 5.7% | 6.2% | 22.7% | 61.3% | 6.00 |
| 4 | Pedersen Mads | 4.9% | 6.4% | 4.5% | 20.9% | 68.7% | 9.00 |
| 5 | Laporte Christophe | 4.6% | 7.4% | 3.9% | 22.5% | 67.6% | 41.00 |
| 6 | De Schuyteneer Steffen | 4.1% | 2.6% | 4.4% | 13.4% | 69.5% | 101.00 |
| 7 | Dainese Alberto | 3.0% | 1.7% | 3.3% | 10.0% | 55.8% | 51.00 |
| 8 | Turner Ben | 2.5% | 3.0% | 2.4% | 13.1% | 58.9% | 41.00 |
| 9 | Groves Kaden | 2.1% | 2.1% | 2.1% | 8.2% | 47.4% | 19.00 |
| 10 | Coquard Bryan | 1.3% | 0.7% | 1.4% | 2.7% | 73.2% | 51.00 |
| 11 | Laurance Axel | 1.2% | 1.4% | 1.2% | 5.7% | 47.7% | 41.00 |
| 12 | Aular Orluis | 1.1% | 1.4% | 1.0% | 6.0% | 45.4% | 51.00 |
| 13 | Brustenga Marc | 1.1% | 0.9% | 1.1% | 5.9% | 55.8% | 251.00 |
| 14 | Pogacar Tadej | 1.0% | 0.7% | 1.1% | 2.6% | 15.1% | 21.00 |
| 15 | Govekar Matevz | 1.0% | 1.7% | 0.8% | 8.6% | 54.2% | 201.00 |
| 16 | Macias Cesar | 0.9% | 1.2% | 0.8% | 6.5% | 54.6% | 251.00 |
| 17 | Hofstetter Hugo | 0.9% | 0.9% | 0.8% | 6.5% | 67.1% | 41.00 |
| 18 | Renard Alexis | 0.6% | 1.0% | 0.6% | 5.5% | 46.8% | 126.00 |
| 19 | Miquel Delgado Pau | 0.6% | 1.1% | 0.5% | 3.3% | 45.4% | 34.00 |
| 20 | González David | 0.6% | 0.2% | 0.7% | 1.4% | 54.9% | 151.00 |
| 21 | Sentjens Sente | 0.4% | 1.4% | 0.2% | 4.2% | 49.2% | 301.00 |
| 22 | Oliveira Ivo | 0.3% | 0.5% | 0.2% | 3.8% | 17.5% | 201.00 |
| 23 | Cort Magnus | 0.3% | 0.1% | 0.3% | 0.8% | 10.6% | 67.00 |
| 24 | Canal Carlos | 0.3% | 0.3% | 0.2% | 1.8% | 23.8% | 301.00 |
| 25 | Nys Thibau | 0.2% | 0.3% | 0.2% | 0.8% | 7.0% | 101.00 |
| 26 | Braet Vito | 0.2% | 0.1% | 0.2% | 0.7% | 42.0% | 81.00 |
| 27 | Albanese Vincenzo | 0.1% | 0.1% | 0.1% | 0.6% | 15.8% | 201.00 |
| 28 | Tronchon Bastien | 0.1% | 0.1% | 0.1% | 1.4% | 14.5% | 201.00 |
| 29 | Garofoli Gianmarco | 0.1% | 0.1% | 0.1% | 0.6% | 7.6% | 401.00 |
| 30 | Hayter Ethan | 0.1% | 0.2% | 0.1% | 0.8% | 12.4% | 126.00 |
| # | Rider | Blend | Cascade | LambdaRank | Podium | Top 15 | Book odds |
|---|---|---|---|---|---|---|---|
| 1 | Brennan Matthew | 46.0% | 40.0% | 45.6% | 58.7% | 78.2% | 1.85 |
| 2 | Meeus Jordi | 11.7% | 11.7% | 11.2% | 32.3% | 64.1% | 13.00 |
| 3 | van Aert Wout | 8.0% | 6.2% | 8.2% | 23.5% | 54.5% | 6.00 |
| 4 | Laporte Christophe | 6.0% | 8.5% | 5.1% | 26.8% | 65.4% | 41.00 |
| 5 | Pedersen Mads | 5.8% | 8.3% | 4.9% | 23.2% | 61.5% | 9.00 |
| 6 | Turner Ben | 4.1% | 3.5% | 4.1% | 14.6% | 55.3% | 41.00 |
| 7 | De Schuyteneer Steffen | 2.5% | 2.4% | 2.4% | 11.4% | 51.8% | 101.00 |
| 8 | Pogacar Tadej | 2.4% | 4.3% | 1.8% | 11.4% | 31.2% | 21.00 |
| 9 | Laurance Axel | 1.9% | 2.4% | 1.8% | 8.2% | 47.6% | 41.00 |
| 10 | Dainese Alberto | 1.6% | 1.0% | 1.7% | 5.4% | 33.8% | 51.00 |
| 11 | Groves Kaden | 1.3% | 1.4% | 1.2% | 5.3% | 30.1% | 19.00 |
| 12 | Aular Orluis | 1.2% | 1.4% | 1.1% | 5.7% | 40.7% | 51.00 |
| 13 | Renard Alexis | 1.1% | 1.6% | 1.0% | 7.1% | 43.6% | 126.00 |
| 14 | Coquard Bryan | 1.1% | 0.4% | 1.2% | 1.2% | 57.9% | 51.00 |
| 15 | Govekar Matevz | 1.0% | 1.8% | 0.8% | 8.3% | 42.5% | 201.00 |
| 16 | Miquel Delgado Pau | 0.9% | 1.1% | 0.7% | 4.3% | 50.4% | 34.00 |
| 17 | Hofstetter Hugo | 0.8% | 1.2% | 0.7% | 6.6% | 52.8% | 41.00 |
| 18 | Sentjens Sente | 0.5% | 1.7% | 0.2% | 4.6% | 35.4% | 301.00 |
| 19 | González David | 0.5% | 0.1% | 0.5% | 1.4% | 47.6% | 151.00 |
| 20 | Canal Carlos | 0.4% | 0.4% | 0.4% | 2.5% | 30.9% | 301.00 |
| 21 | Brustenga Marc | 0.3% | 0.1% | 0.3% | 2.4% | 29.5% | 251.00 |
| 22 | Cort Magnus | 0.3% | 0.2% | 0.3% | 0.7% | 7.6% | 67.00 |
| 23 | Braet Vito | 0.2% | 0.1% | 0.3% | 1.0% | 43.7% | 81.00 |
| 24 | Macias Cesar | 0.2% | 0.2% | 0.2% | 2.8% | 28.9% | 251.00 |
| 25 | Mulubrhan Henok | 0.2% | 0.1% | 0.2% | 0.8% | 9.5% | 201.00 |
| 26 | Nys Thibau | 0.2% | 0.2% | 0.2% | 0.7% | 7.1% | 101.00 |
| 27 | Fisher-Black Finn | 0.2% | 0.2% | 0.2% | 1.8% | 9.1% | 67.00 |
| 28 | Garofoli Gianmarco | 0.2% | 0.2% | 0.2% | 0.9% | 12.5% | 401.00 |
| 29 | Oliveira Ivo | 0.2% | 0.2% | 0.2% | 1.4% | 11.9% | 201.00 |
| 30 | Albanese Vincenzo | 0.2% | 0.2% | 0.2% | 0.8% | 18.3% | 201.00 |
After the stage
| Pos | Rider | Team | Versatile sprint win p | Pauls sticks win p | Full sprint win p | Hindsight (post-hoc) win p |
|---|---|---|---|---|---|---|
| 1 | Brennan Matthew | Team Visma | Lease a Bike | 45.6% | 26.2% | 43.0% | 46.0% |
| 2 | Meeus Jordi | Red Bull-BORA-hansgrohe | 10.0% | 2.9% | 16.6% | 11.7% |
| 3 | Albanese Vincenzo | EF Education-EasyPost | 0.2% | 0.3% | 0.1% | 0.2% |
| 4 | Romele Alessandro | XDS Astana Team | 0.0% | 0.0% | 0.0% | 0.0% |
| 5 | Laurance Axel | Netcompany INEOS Cycling Team | 2.1% | 4.1% | 1.2% | 1.9% |
| 6 | Laporte Christophe | Team Visma | Lease a Bike | 5.1% | 9.2% | 4.6% | 6.0% |
| 7 | Dainese Alberto | Soudal Quick-Step | 1.4% | 0.2% | 3.0% | 1.6% |
| 8 | Braet Vito | Lotto-Intermarché | 0.2% | 0.2% | 0.2% | 0.2% |
| 9 | González David | Pinarello-Q36.5 Pro Cycling Team | 0.5% | 0.3% | 0.6% | 0.5% |
| 10 | van Aert Wout | Team Visma | Lease a Bike | 8.4% | 6.1% | 6.1% | 8.0% |
| Scenario | Blend win p | Book devig |
|---|---|---|
| Versatile sprint | 45.55% | 35.87% |
| Pauls sticks | 26.17% | 35.87% |
| Full sprint | 43.03% | 35.87% |
| Hindsight (post-hoc) | 46.04% | 35.87% |
Reflection
Brennan won again, and the base read called the finishing order at the top exactly: 'versatile sprint' had him #1 at 47.7% against a book that already made him a 1.85 favourite (35.9% devig), and Jordi Meeus #2 at 9.9% against 5.1% — they finished first and second. Two stage wins in four racing days for a twenty-year-old, and the model has been ahead of the market on him both times. Behind them the board stayed useful rather than perfect: Laporte #5 finished 6th, Laurance #8 finished 5th, Dainese #9 finished 7th, van Aert #3 finished 10th.
The shape revision was right, and it is worth being precise about how right. I keyed 0.35 on `bunch` against the runbook's stage-2 policy of pure-composition keying, on the argument that stage 5 was easier than stage 2 in every dimension that drives group size. The road agreed emphatically: 89 riders finished on the winning time, 48.9% of the field — not merely over the 20% macro bar but two and a half times stage 2's 22.3%. Alto de Pauls cost the bunch essentially nothing. Tier-3 records exact=False, covered=True, weight_on_true=0.35, which is partial credit where stage 2's composition-pure key scored zero. Campaign record now 2/4 exact, 3/4 covered, mean weight 0.49. If anything the honest verdict is that 0.35 was still too timid — but the composition half was also vindicated, since the winner was a versatile fast man and not a pure sprinter, so the mixed key was reading the day correctly on both axes and merely under-weighting one.
The post-hoc run says, for the third stage running, that knowing the true shape would not have helped. Rebuilt from five days whose winner-group PROPORTION matched the actual 48.9% — deliberately disjoint references from the pre-race set — the hindsight board is nearly identical at the top (Brennan 45.6% vs 47.7%, Meeus 11.2% vs 9.9%) and found none of the surprises: Albanese, 3rd at 201.0, moved from #29 to #31; Romele, 4th at 81.0, stayed at #63. Pool retrieval got WORSE, 7 of the actual top 15 against the pre-race board's 8. That is now a robust campaign finding rather than an anecdote — stages 4 and 5 both say the same thing the Tour's stage 17 said, that reference-matching on how a day ENDED imports a selection story without identifying the individuals who were there.
The retrieval channel had its best day yet, and by the same mechanism as stage 2. The day-menu board held 8 of the actual top 15 against the production board's 7: it promoted Pau Miquel, who finished 11th, and dropped Pogacar, who finished 49th. That is the second time this race the channel's promotion has been Pau Miquel and the second time he has justified it — 2nd in Manosque, 11th here. Across the four scored stages the ledger is now stage 2 −1, stage 4 level (+1 on the break board), stage 5 +1: roughly neutral overall, with the wins concentrated exactly where theory says they should be, on days whose key is a composition the production ranking under-weights.
Two names to carry forward, in opposite directions. Mads Pedersen was the model's #4 (4.7%) and the book's third choice (7.4%) and finished 156th; Ben Turner was #6 and finished 171st. Both simply stopped — a bunch that keeps half the field together does not shed riders to 156th by attrition, so this reads as riders sitting up rather than being dropped, which no board can price. Against that, Christophe Laporte was the flagged watch item after Manosque (#3 on the board, 57th on the road) and this time delivered 6th from #5, so that particular over-ranking concern is answered. The open question the stage-5 result sharpens is Pedersen's condition, not the model's read of him.