Pacing
Where the day was actually won and lost
Your start pen costs you 7 min 59 s
Riders are seeded into 17 pens by expected finish time, the first leaving at 06:30 and each one 7.5 minutes later — a two-hour spread across the field. Comparing raw finish times across pens would just rediscover that seeding, so this doesn’t do that. Instead every one of the 12,165 riders here is matched against others of measured identical climbing ability (their combined time up the 3 timed cols, in 10-minute bands), and then compared only on the parts of the day that were not used to match them.
Each bar is one start pen. Above the line, that pen rode the non-climbing parts of the day slower than riders of identical measured climb ability; below it, faster. Hover a pen for its numbers.
start pen — SAS 0 leaves at 06:30, each one 7.5 min later
The same comparison, split by section: how much longer the back pens took than equally-strong riders from the front pens.
Two things this measurement cannot tell you
- The back pens lose their field. 15% of pen 0 failed to finish; in the last four pens it is 41%. Whoever is left back there is a survivor, which is the most likely reason the penalty stops growing — and even reverses — in the very last pens.
- This is not a randomised trial. Riders are seeded, not assigned by coin flip, so two people matched on climbing time may still differ in ways the cols don’t capture. The control below is what makes the result credible, not the matching alone.
The control that makes this hold up
Col de Sarenne is a climb, and it was deliberately left out of the matching so it could be used as a check. If the back pens were simply weaker riders, they would be slower here too. They are 1 min 22 s faster — the effect reverses on the one climb we can test it on. Whatever is costing the back pens their time, it is not their legs on a mountain. It is what happens on the roads in between.
When the race stops being decidable
At every checkpoint each finisher is ranked against the other finishers who reached it, and compared with where they ended up. The gap between the two is how much of the field they still had left to move through — from 10.09 points at Pied de Col de la Croix de Fer down to 0.36 points by Col de Sarenne. Ranks at a checkpoint count riders who later abandoned, which would credit everyone with the places those riders vacated, so they are re-ranked among finishers only. Done that way the finish line comes out at exactly zero, which is the check that the arithmetic is right.
The typical rider's remaining move, in places of the field, between each checkpoint and the finish line of L'Étape du Tour 2026. Hover a checkpoint for its numbers.
How much of the race was still undecided, checkpoint by checkpoint. Use the left and right arrow keys to read each point.
The last checkpoint must fade to zero: by the finish line there is nothing left to move. That is the arithmetic self-test, not a finding.
Show the numbers
| Checkpoint | Median remaining move |
|---|---|
| km 15 | 10.1 pt |
| km 39 | 6.4 pt |
| km 80 | 5.5 pt |
| km 92 | 4.2 pt |
| km 97 | 4.2 pt |
| km 115 | 2.1 pt |
| km 115 | 2.1 pt |
| km 116 | 2.0 pt |
| km 140 | 1.6 pt |
| km 161 | 0.4 pt |
| km 170 | 0 pt |
The line does not fall smoothly, and the bumps are the interesting part: position at the foot of a climb is worth much less than position at its summit. The field arrives at a col in whatever groups the valley assembled, and then the mountain sorts it out again.
the climbs do 83% of the sorting
The obvious way to ask this — take riders within a couple of minutes of each other and see how far apart they finish — turns out to measure the wrong thing. Early on the whole field is packed into a few minutes, so being level on the clock at Pied de Col de la Croix de Fer means something completely different from being level on the clock at the top of a col. So this conditions on position instead: among riders sharing your place in the field at a checkpoint, how widely scattered are their finishes?
At Pied de Col de la Croix de Fer the answer is 43.3 points of the field — knowing someone’s position there tells you very nearly nothing about where they will finish. Every section after it closes that number down.
Among riders sharing your position at a checkpoint, the spread of where they finished. Each step down is a section of road doing the sorting; the colour is the terrain of the section that got you there.
How much was still open at each timing mat. Use the left and right arrow keys to read each point.
Show the numbers
| Checkpoint | Still open |
|---|---|
| km 15 | 43.3 pt |
| km 39 | 28.9 pt |
| km 80 | 23.4 pt |
| km 92 | 17.9 pt |
| km 97 | 18.5 pt |
| km 115 | 7.9 pt |
| km 115 | 7.5 pt |
| km 116 | 7.3 pt |
| km 140 | 5.3 pt |
| km 161 | 0.3 pt |
| km 170 | 0 pt |
| Section ending at | Terrain | Still open | Settled here |
|---|---|---|---|
| Pied de Col de la Croix de Fer · km 14.8 | flat | 43.3 pt | — |
| Col de la Croix de Fer · km 38.5 | climb | 28.85 pt | 33% |
| Pied de Col du Télégraphe · km 80.1 | flat | 23.35 pt | 19% |
| Col du Télégraphe · km 92 | climb | 17.94 pt | 23% |
| Pied de Col du Galibier · km 97.3 | flat | 18.53 pt | 0% |
| Tunnel du Galibier · km 114.6 | climb | 7.87 pt | 58% |
| Activation Canyon · km 115 | climb | 7.52 pt | 4% |
| Col du Galibier · km 115.6 | climb | 7.25 pt | 4% |
| Pied de Col de Sarenne · km 140.2 | descent | 5.3 pt | 27% |
| Col de Sarenne · km 161 | climb | 0.31 pt | 94% |
| FINISH · km 170 | descent | 0 pt | 100% |
Read it as a rider and it says something useful: your position at the bottom of a climb is worth far less than your position at the top. Across this course the spread at a summit is 11.87 points against 22.53 points on the approaches. You arrive at a col in whatever group the valley handed you; the climb then throws that away and sorts everyone on their own legs.
Two ways to read this wrong
- Earlier sections have more to take. The first climb starts with almost all the uncertainty still on the table, so it will always look productive in raw points. The “settled here” column is the share of what was still open at that moment, which is the fairer comparison — by that measure it is FINISH, taking 100% of what remained.
- The last section is flattered by being last. Whatever is unresolved when you reach the final mat gets resolved there by definition, so the terrain totals above are in points rather than shares.
12,733 classified finishers · built 2026-09-06 · back to the explorer