Omnium medal odds
After the scratch, the tempo and the elimination — the three moments in an omnium when the standings are certain — what is the probability that a rider in a given position, with a given cushion or deficit to the medal line, ends up on the podium? Fitted on 91 omniums from 2017–2026.
One omnium, probability by stage
Each line is a rider's modelled chance (of a medal, or of the win — switch above) after the scratch, the tempo and the elimination — the three points at which the standings are known for certain — and then the result. The points race is deliberately not sliced up: the sheets don't say when laps are taken, so a mid-race number would be a guess dressed as a prediction. Final medallists are coloured and named.
| Rider | After scratch | After tempo | After elimination | Final |
|---|
Where do you stand?
Set the stage and your position, then two margins, used separately: the medal-line margin (inside the top three: points over 4th; outside: points behind 3rd) drives the medal probability, and the gold margin (leading: points over 2nd; otherwise points behind the leader) drives the win probability.
Rider specialities
Where does each rider make their omnium? Average placing in each of the four races across every omnium in the chosen years (riders with at least the minimum number), where they typically stand before the points race, and where they finish. Add a second and third rider to compare profiles side by side. A rider who finishes higher than they stand before the points race is a closer; one who finishes lower has been living off the bunch races.
Each dot is a rider's average over their omniums: standing after the elimination (x) against final position (y). Below the diagonal: finishes better than they stood — the points race is their race. Above it: fades. Dot size = number of omniums. Click a dot or pick a rider to see their profile; sort the table by clicking a heading.
| Rider | Omniums | Scratch | Tempo | Elimination | Points race | Before points race | Final | Climb | Medals |
|---|
How the probability is calculated
Four steps, drawn with the real numbers from the model — the first is the summary; the rest is the recipe.
1 · Slice every omnium into states. At each of the three checkpoints — after the scratch, after the tempo and after the elimination — the standings are frozen. Each rider at each checkpoint is one row: 91 omniums × ~20 riders × 3 checkpoints ≈ 5,700 rows, each tagged with what happened in the end (medal or not, gold or not).
2 · Describe each state with three numbers. The margin to the line that matters (medal: points over 4th if inside the top three, points behind 3rd if not; gold: points over 2nd if leading, points behind the leader if not), the log of position (so 1st→2nd counts for more than 9th→10th), and an inside flag (1 if you currently hold a medal place / the lead).
The evidence
What actually happened, before any modelling: the share of riders in each position after each race who finished on the podium (or won — switch above), and the same by points margin after the elimination.
Model
If you want to know how the probabilities are produced, get in touch — the tools are free; doing your selection isn't.
Limits. The points race is not modelled mid-race because lap gains are not timed on the sheets. Field strength, nationality, the rider's own history and the venue are not features — the model is deliberately about race state. Probabilities are what happened in 2017–2026 omniums to riders in that state, nothing more.