Cycling race predictor · Okolo Slovenska

Stage 1 — 2026-09-16

The same production stack priced under 1 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.

1 scenarios 130 riders priced models retrained d911de60f7 book snapshot no capture

The fork

1 readings of the same road

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.

Hindsight (post-hoc)

Hindsight (post-hoc) — oracle-matched references
sprint0.77
mountain0.03
hilly0.20
rolling0.01
break0.00
TT0.00
keyed: bunch sprint (board order by rider-type retrieval; prices unchanged)
RETROSPECTIVE ONLY — no pre-race prediction was made. This board uses the settled result as the oracle and the nearest historical realized race shapes. It is corpus completion, not evidence of forecasting skill.
Top 5 (blend): Magnier Paul 63.3%; Milan Matteo 5.2%; Stolic Mihajlo 1.4%; Bittman Adam 0.8%; Biermans Jenthe 0.7%

The board

Win probabilities, scenario by scenario

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 Hindsight (post-hoc) 78% — 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.

RiderTeamHindsight (post-hoc)Book oddsBook devigMax edge
Magnier PaulSoudal Quick-Step63.3%n/a––
Milan MatteoGroupama-FDJ United5.2%n/a––
Stolic MihajloBahrain Victorious Development Team1.4%n/a––
Hindsight (post-hoc) — oracle-matched references — top 30, all heads
#RiderBlendCascade LambdaRankPodiumTop 15 Book odds
1Magnier Paul63.3%46.9%85.2%66.1%82.3%n/a
2Milan Matteo5.2%8.7%5.7%25.0%58.2%n/a
3Stolic Mihajlo1.4%2.1%1.6%9.7%44.4%n/a
4Bittman Adam0.8%1.7%0.9%5.0%32.3%n/a
5Biermans Jenthe0.7%2.4%0.5%6.4%34.8%n/a
6Walls Matthew0.6%1.0%0.7%4.2%33.1%n/a
7Neuman Dominik0.5%0.8%0.5%4.2%40.2%n/a
8Capra Thomas0.4%1.1%0.3%4.5%32.2%n/a
9Källberg Axel0.3%0.6%0.3%4.1%40.5%n/a
10Bravo Henrique0.2%0.8%0.2%3.1%23.2%n/a
11Evertsen-Hegreberg Simen0.2%0.6%0.2%5.5%41.7%n/a
12Haluska Martin0.2%0.8%0.1%5.4%28.2%n/a
13Barthe Cyril0.2%0.6%0.2%3.5%30.8%n/a
14Villa Giacomo0.2%0.6%0.2%3.4%36.2%n/a
15Greenwell Samuel0.2%0.5%0.2%3.6%28.1%n/a
16Agnoletto Blake0.2%0.6%0.2%4.4%25.2%n/a
17Consolidani Leonardo0.2%0.6%0.1%3.8%27.8%n/a
18Terrier Lucas0.2%0.5%0.1%3.9%30.2%n/a
19Dubois Yann0.1%0.3%0.1%1.0%24.2%n/a
20Novak Samuel0.1%0.1%0.1%1.0%15.9%n/a
21Vanicek Simon0.1%0.3%0.1%0.9%12.0%n/a
22Ilski Szymon0.1%0.4%0.1%0.8%15.6%n/a
23Kjeldsand Karl Emil0.1%0.2%0.1%0.6%14.8%n/a
24Slock Liam0.1%0.2%0.1%0.8%11.2%n/a
25Cisneros Said0.1%0.3%0.1%0.9%8.0%n/a
26Pekala Piotr0.1%0.1%0.1%0.6%12.1%n/a
27Kull Robin0.1%0.1%0.1%0.7%11.8%n/a
28Quadriglia Davide0.1%0.3%0.0%0.7%16.0%n/a
29Fortin Filippo0.1%0.3%0.0%0.5%5.8%n/a
30van Gestel Dries0.1%0.2%0.1%3.3%11.4%n/a

After the stage

Results — won by Magnier Paul

PosRiderTeamHindsight (post-hoc) win p
1Magnier PaulSoudal Quick-Step63.3%
2Biermans JentheCofidis0.7%
3Granberg SanderTeam Drali-Repsol0.0%
4Villa GiacomoBiesse-Carrera Premac0.2%
5Milan MatteoGroupama-FDJ United5.2%
6Capra ThomasBahrain Victorious Development Team0.4%
7Fortin FilippoHrinkow Advarics0.1%
8Källberg AxelLucky Sport Cycling Team0.3%
9Neuman DominikATT Investments0.5%
10Slock LiamLotto-Intermarché0.1%

4 riders in the top 15 are shown under a provisional identity pending reconciliation: PAKOT András Gusztáv, KADLEC Milan, PAKOT András Gusztáv, KADLEC Milan.

Scenario scoring — probability given to the actual winner

ScenarioBlend win pBook devig
Hindsight (post-hoc)63.28%–