Grand tour team selection: how eight places get filled

A squad announcement looks like a list of the eight best riders a team could find. It never is. It is the answer to a constrained optimisation problem, worked out over months, and the interesting part of grand tour team selection is everything that had to be left out.

Twenty-one race days. Two rest days. Flat sprints, transition stages, a time trial or two, high mountains, and a final week where the race is decided by riders who are all deeply tired. Eight places to cover it, and no way to add a ninth once the race begins.

That constraint is what turns selection from a ranking exercise into a portfolio exercise. Nobody is choosing the eight best riders. They are choosing eight riders whose combined coverage of twenty-one different problems maximises the chance of the specific outcome the team has decided it wants, weighted by the chance that two of them will be in an ambulance by the second Tuesday.

Grand tour team selection is a portfolio problem, not a ranking

The instinct is to rank the roster and take the top eight. Every directeur sportif has watched somebody do that and watched it fail.

Here is why. Suppose a team has three genuinely excellent climbers. In a ranking, all three go. In a race, only one of them can be the leader, and the other two spend three weeks doing a job that requires a slightly different quality: not the ability to win a mountain stage, but the ability to ride at a specific hard pace for a specific twenty minutes at a specific point in a stage, having already done four hours of work, and then stop. Excellent climbers are frequently bad at that, because the instinct that makes someone a good climber is to respond when the pace lifts, and a domestique's job is to ignore the pace lifting and keep riding their own.

Now suppose that team has no rider who can hold position in a crosswind. The first flat stage on an exposed coast splits the race, their leader is on the wrong side of the split, and three weeks of climbing preparation is worth nothing by lunchtime on day four. The best eight riders lost to a squad with a worse third climber and two riders who know how to fight for wheels.

Selection is about coverage, and coverage is about the worst case rather than the best. A grand tour squad is picked the way you would pack for a long expedition: not by taking the finest individual pieces of equipment, but by making sure that no single foreseeable event leaves you without something you need.

There is a second reason ranking fails, and it is less obvious. The value of a rider in a squad depends on the other riders in it. A pure sprinter is worth a great deal with a lead-out and very little without one, so the decision to take him is really a decision to take two or three riders. A climber's value depends on whether there is someone to hold position for him on flat days. This is why squads tend to arrive in clusters rather than individually, and why an argument about the last place is usually an argument about a group of three.

The final complication is that the team is choosing under uncertainty about its own riders. Form is estimated from training data, from a build-up race, and from what a rider says about how they feel, which is not always what they mean. A squad picked in April is a set of forecasts. A squad confirmed the day before the start is a set of slightly better forecasts.

Eight places, twenty-one days, and no way to change your mind

The rules do most of the constraining before anyone gets to the interesting decisions.

• 8Riders each team may start with
• 23Teams in a men's grand tour field
• 21Race days in a grand tour
• 0Substitutions permitted once the race has started

Figures set by the UCI regulations and the grand tour format rather than measurements. Team size at grand tours was reduced from nine to eight from 2018, and the UCI Management Committee raised the men's grand tour field from 22 teams to 23.

Eight riders is the number that shapes everything else. It was nine until 2018, and the reduction was made on safety grounds, to shrink the size of the peloton and the convoy behind it. The effect on tactics was larger than the effect on safety, and largely unintended.

With nine, a strong team could control a stage on its own: put six riders on the front, ride a pace nobody can attack, and deliver a leader to the final climb without ever being troubled. With eight, that gets marginally harder, and the marginal difference matters because it sits right at the point where control becomes possible. Take a sprinter and a lead-out out of nine and you still have six for the mountains. Take them out of eight and you have five, and five is not quite enough to hold a race closed.

The 23-team field is a newer change and cuts the other way. More teams means more riders wanting a stage win, more squads with nothing to lose, and more competition to get into a breakaway. A field with an extra team is a field where the day's escape is harder to form and harder to control once it has gone, which raises the cost of chasing and therefore the number of days a big team decides not to bother.

The no-substitution rule is the one that does the most quiet damage. Before the start, teams travel with reserves and can swap right up to the organiser's confirmation deadline, which is why a squad announced two weeks out is provisional and why the final list sometimes appears the day before the race. After the start, a rider who abandons is simply gone. Illness runs through a team bus. A crash in the first week takes two riders at once. There is no mechanism for repair, and the mountains do not get easier to cover because you now have five riders instead of seven.

Every experienced selector therefore builds in redundancy they hope to waste. That is a genuinely uncomfortable thing to do, because redundancy is invisible when it works.

What the eight places actually have to cover

Strip the job titles away and a grand tour squad has to be able to do a fixed set of things.

Somebody has to finish high on general classification, if that is the objective. Somebody has to be able to ride on the front on a flat stage for two hours without it costing them the next day. Somebody has to be able to fight for position in the last fifteen kilometres before a narrow road, a crosswind section or a summit finish, which is a different skill from riding fast. Somebody has to be able to go up a mountain quickly enough to hurt other people's leaders. Somebody has to be able to get to the front and stay there in the final kilometre if the team wants sprint results. Somebody has to be able to go back to the car, collect eight bottles, and carry them forward through a peloton at fifty kilometres an hour, repeatedly, in heat, without complaint. Somebody has to make decisions.

13%38%25%13%13%

• Leader for the general classification1
• Mountain domestiques who can climb late and hard3
• Rouleurs for the flat, the wind and the front of the bunch2
• Road captain and positioning specialist1
• Free role, for the breakaway and the day the plan fails1

A constructed illustration of one plausible allocation, not any team's real squad. The point is that the eight sum to eight, so every place given to one function is a place taken from another.

Show the numbers

Invented worked example: eight places in a squad built around the overall

Item	Value
Leader for the general classification	1
Mountain domestiques who can climb late and hard	3
Rouleurs for the flat, the wind and the front of the bunch	2
Road captain and positioning specialist	1
Free role, for the breakaway and the day the plan fails	1

Look at what that allocation does not contain. No sprinter. No lead-out. No dedicated time trial rider, because the leader is assumed to handle the time trial himself and the rouleurs can hold a decent pace on a flat day. That squad can win the overall and will probably not win a single stage unless the leader wins one or the free role gets lucky.

Now the same eight places, allocated by a team that wants stage wins as well.

• Overall-first squad
• Split-objective squad

Leaders12

Mountain support31

Flat rouleurs21

Sprint lead-out02

Road captain and free role22

Constructed illustration comparing two hypothetical allocations of the same eight places. Neither is any team's real selection. The totals are eight in both cases, which is the whole point.

Show the numbers

Invented worked example: the cost of adding a sprinter to the same eight places

Item	Overall-first squad	Split-objective squad
Leaders	1	2
Mountain support	3	1
Flat rouleurs	2	1
Sprint lead-out	0	2
Road captain and free role	2	2

The split-objective squad has cut its mountain support from three riders to one. That is not a small adjustment. On the day the race is decided, its leader arrives at the foot of the final climb alone, having spent energy positioning himself, while the overall-first team still has two riders to set the pace and a third who was there ten minutes ago.

Whether that trade is correct depends entirely on what the team is being paid to achieve, which is a question about sponsorship and standing rather than about bike racing. Both squads above are defensible. Neither is a mistake. They are answers to different questions.

One leader or two, and the cost of not deciding

Backing one rider is clean and fragile. Backing two is robust and expensive.

The case for a single leader is that a bike race is decided by a small number of moments where somebody has to commit, and hesitation costs more than a slightly weaker rider does. If two riders are protected, the team cannot ride the front for either of them without partially wasting the effort, and the moment one attacks, the other's team-mates are chasing their own colleague. Riders can feel that ambiguity from inside the race and it changes what they are willing to spend.

The case for two is arithmetic. A grand tour is three weeks long and the ways to lose one are numerous: a crash on a wet descent, a stomach bug, a mechanical at the wrong moment, one bad day at altitude, a positive test on a team-mate that unsettles everyone. Put all the value in one rider and there is a real probability the whole three weeks yields nothing. Two leaders halve that risk at the cost of diluting the support each receives.

A third structure exists and is the one that actually appears most often: a stated leader and an unstated second option who is given more freedom than a domestique and less protection than a leader. That arrangement lets the team keep its options without saying so, and it works until the second rider is ahead of the first on the road, at which point somebody has to make a decision in a radio conversation with two very tired men listening.

The honest version of the argument is that co-leadership is a hedge, and hedges cost money. Teams under pressure to deliver a result hedge. Teams confident in one rider do not. The pattern is visible in the way squads are announced: a team that says nothing about hierarchy is a team that has not decided, and a team that has not decided will decide in week two, badly, in public.

There is one situation where two leaders is unambiguously right, and it is when the two are good at different things. A climber and a rider who is strong against the clock are not competing for the same days. The team's plan is then legible: protect both to the mountains, and let the road sort out which one is in a position to use the time trial. The way those two profiles score across a route is worked through in more detail in the piece on how the jerseys and classifications shape a race.

Why the last man on the climb matters more than the strongest rider on paper

This is the part of selection that most people get wrong, and it is the reason a squad list can look weak and be excellent.

Picture the decisive climb of the decisive stage. The leaders are together. Each has whatever support survived the previous four hours. The pace on the front is being set by one rider, and that rider is not going to win anything, because he will be dropped in a few minutes and finish twenty minutes down.

His job is to ride at a pace that is uncomfortable for everyone behind him. Not a pace that drops people, particularly. A pace that makes attacking expensive, that discourages the opportunists, that keeps the group moving fast enough that nobody gains anything by sitting up. He does that until he cannot, and then the next rider does it, and the leader who still has a team-mate at the bottom of the last climb has an advantage over a leader who does not.

The value of that rider is not measured by how good a climber he is in absolute terms. It is measured by his power at a specific point in a specific stage after specific accumulated fatigue, at altitude, in whatever weather turned up. A rider who could beat him in a one-hour hill climb test might be useless in that role, because the test does not include the four hours before it.

Three qualities decide it. Durability, meaning how much of a rider's fresh capacity survives a long day and then a long week. Altitude tolerance, which varies enormously between individuals and is not predictable from sea-level numbers. And a temperament that can ride at a fixed effort while a race happens around them, without responding to attacks that are not their problem.

None of those three shows up in a results list, which is why the riders who fill these places are frequently anonymous to the public and precisely valued inside the sport. It is also why teams recruit them from altitude camps and training data rather than from race wins, and why the metrics used to identify them are power-based rather than results-based, a subject covered in the piece on what power output per kilogram actually tells you.

The practical rule that falls out of this: a squad's climbing strength is not the sum of its climbers, it is the quality of its third-best climber late in a stage. The best climber is the leader and is being saved. The second is doing the middle of the climb. It is the third who determines whether the leader is still protected when it matters.

The sprinter question is the whole selection in miniature

Nothing exposes the trade-offs like the decision about whether to bring a fast finisher at all.

The case for is straightforward. A grand tour contains flat stages, and a flat stage is the single most reliable opportunity in the race to win something. A breakaway win is a lottery. A mountain stage win requires being one of the best climbers in the world on that day. A bunch sprint is a repeatable event where a genuinely quick rider with an organised team has a real chance several times over three weeks. Sponsors count stage wins and television time, and a sprinter delivers both in a way that a rider finishing sixth overall does not.

The case against is the accounting. A sprinter cannot be brought alone. He needs at least one rider to deliver him into the final kilometre and usually two, plus somebody willing to chase a breakaway on a day the team wants a sprint, which is one of the least rewarding jobs in cycling. And the cost does not end when the road goes uphill.

On mountain days a sprinter has to finish inside the time limit, calculated as a percentage of the stage winner's time and varying with the difficulty and average speed of the stage. Missing it means elimination. So the group of sprinters and their helpers forms at the back and rides its own race against the clock, and that group needs at least one competent rider from each team to manage the pacing, the feeding and the arithmetic. The mechanics of that calculation, and how close it routinely gets, are set out in the piece on how the time limit is worked out and who it eliminates.

Count the true cost of a sprinter honestly. One place for him. One or two for the lead-out, whose value on a mountain stage is close to zero. Some fraction of another rider's three weeks spent shepherding him over cols. Against a squad of eight, that is somewhere between a quarter and a third of the team committed to an objective that is decided in the last two hundred metres of about six days.

Teams resolve this in three ways. The largest budgets take both, accept the dilution, and rely on having better riders than everyone else. Teams with a genuine overall contender and no obligation to entertain leave the sprinter at home and go to a different race with him. And a middle group takes a fast rider who is not a pure sprinter: someone who can win from a reduced group after a hard finale, contest intermediate sprints, and still climb well enough to be useful. That rider wins fewer bunch sprints and costs one place instead of three, which is often the better trade. The organisation required to launch a genuine sprinter is described in the piece on how a lead-out train is actually run.

The time trial specialist, and the awkwardness of a single day

A route with a long individual time trial rewards a particular kind of rider, and here selection runs into a problem of shape.

A time trial specialist is usually heavier, more powerful and worse at climbing than the riders around him. On the day of the test he is enormously valuable. On the other twenty days he is a rider who cannot follow on a mountain, which makes him an expensive luxury in a squad of eight unless he can do something else.

So the question is never really whether to take a time trial specialist. It is whether the leader is good enough against the clock, and if not, whether the gap can be covered by a rider who is useful the rest of the time. A powerful rouleur who can hold a fast pace on the flat, protect the leader in crosswinds, ride the front for two hours and also produce a respectable time trial is worth more than a pure specialist, even though he would lose to the specialist in a straight test.

The route's time trial content changes the whole calculation. A course with a substantial flat test against the clock favours a heavier, more powerful overall contender and reduces the number of climbing places needed, because time can be taken back without a mountain. A route with almost no time trialling makes the race a pure climbing contest, which means the leader must be a climber and the support must be climbers, and a squad built for a different route is simply in the wrong race. The variables that decide how much time a route can actually produce against the clock are covered in the piece on what makes one time trial course faster than another.

There is one more wrinkle. If the route opens with a short prologue or a team time trial, the early days matter for jersey purposes and for the ordering of the peloton, and a squad may take a rider specifically for that. Selecting one of eight places for a single early day is only defensible if the jersey has commercial value to the team, which for some sponsors it plainly does.

The road captain is a selection, not a job title

Race radios carry information but not judgement, and the team car is behind the convoy, watching a television feed and a set of GPS dots, several hundred metres and thirty seconds behind reality.

Somebody in the bunch has to make decisions at the speed the race is actually moving. Which move is dangerous and which is not. Whether to spend two riders now or hold them for the next section. Whether the exposed stretch after the next village is worth fighting for, which requires knowing that it is exposed before arriving at it. When to tell the leader that the plan has changed and he is not going to win today.

That is the road captain, and the role is filled by picking a rider for it rather than by appointing one from among the riders you already picked. The qualities are experience, a good memory for roads, the standing to be obeyed by younger riders, and enough remaining physical capacity to be near the front when the decisions have to be made. A road captain riding at the back is not a road captain.

The awkward part is that the role usually goes to a rider whose own career is past its results-producing phase, which means selecting a road captain is explicitly selecting somebody who will not win anything. Teams with young squads and no such rider tend to discover the value of the role by getting the decisions wrong, expensively, on a windy Wednesday. How that in-race authority interacts with the directeur sportif's instructions is one of the themes of the piece on how teams actually control a race.

Selecting for the crashes you cannot predict

Since there are no substitutes, contingency has to be built into the eight places themselves.

The first form of contingency is overlapping capability. Two riders who can each set pace on a climb is better than one rider who is better at it, because one of them will be sick. Selectors talk about this as having no single point of failure, and it costs performance in the median case to buy insurance in the tail. That is a genuine cost and the reason some teams refuse to pay it.

The second is picking riders who can absorb damage. Crashes in the first week are close to certain for somebody in every team, and there is a real difference between a rider who continues with a wound dressed and one whose three weeks is over. Some of that is luck, some is bike-handling, some is position in the bunch, and some is a willingness to keep going that varies between people in ways teams know about and never say publicly.

The third is illness planning, which is the least glamorous and most consequential. A grand tour team lives in close quarters for three weeks, often in modest hotels, moving daily. Respiratory illness moves through a bus quickly. Teams run their own hygiene protocols, separate riders at the first symptom, and accept that a squad may lose two riders in a week to something nobody crashed into. Selecting one rider who is known to be robust over a marginally stronger one who is not is a defensible decision that looks foolish if the robust rider is simply worse and nobody gets ill.

The fourth is what happens after the losses. A team that plans only for the intended race has no answer to being down to five riders in the Alps. Teams that plan properly have a fallback objective ready: switch from defending a general classification position to hunting stage wins, put riders in breakaways, salvage the three weeks in a different currency. That switch is much easier if at least one rider in the squad was selected partly for his ability to get into a break, which is why the free role in the earlier allocation is not a luxury.

How a squad actually gets picked, from route to start line

The sequence is more or less the same everywhere, and it starts long before anyone talks about riders.

1. 1The route is publishedMonths out, the organiser releases the parcours. Teams count summit finishes, kilometres against the clock, exposed and technical sections, and the placement of the hardest days relative to the rest days.
2. 2The objective is setOverall contention, stage wins, a jersey, or visibility for a sponsor. Everything downstream follows from this, and teams that skip it end up with squads that contradict themselves.
3. 3The leader is chosen, or the decision to have two is takenThis fixes the shape of the rest of the squad, because a climbing leader and a leader who is strong against the clock need different support.
4. 4The non-negotiables are placedThe riders whose specific ability the plan depends on. Usually two or three names that were never in doubt.
5. 5The sprint decision is madeTake a sprinter and commit two or three places to him, take a fast finisher who costs one, or take neither. This is the largest single trade in the selection.
6. 6The remaining places are argued overPositioning, crosswind ability, breakaway capability and a road captain compete for what is left. This is where selection meetings actually spend their time.
7. 7A long list goes to the ridersRiders are told they are in contention, which affects their race programme, their altitude camp and their build-up. Being on the long list is not being selected.
8. 8Form and health are checked lateTraining data, a build-up race, illness, a crash. Any of these can remove a rider a fortnight before the start.
9. 9Reserves travelTeams bring more riders than they can start, in case of illness or a late crash. The reserves train with the squad and prepare as though they are racing.
10. 10The confirmation deadline passesSubstitutions are possible up to the organiser's deadline before the start. After it, the eight are the eight for three weeks.

A structural account of the process rather than any team's internal procedure. The confirmation deadline and the number of reserves permitted are set by the organiser and the UCI regulations for each race.

Two features of that sequence are worth pausing on.

The objective comes second, immediately after the route, and before any rider is discussed. Squads that look incoherent almost always come from teams that reversed those steps, picked their best available riders, and then tried to work out what to do with them.

And the long list is a real commitment even though it is not a selection. A rider told he is in contention structures his entire spring around it: altitude camp, race programme, weight, recovery. Being dropped from a long list two weeks before a grand tour costs a rider his season, not his month, which is why these conversations are handled carefully and why teams over-communicate through the process.

Why a team might not select for the general classification at all

Roughly half a grand tour field is not trying to win it, and their selections are frequently more interesting than those of the teams that are.

A team with no realistic overall contender has a clear-eyed choice. Chasing eighth place overall requires most of the squad to ride conservatively for three weeks and produces a result that no sponsor can use, no highlights package includes and nobody outside the sport can name. Chasing stage wins requires riders who can get into breakaways, survive to the finish of a hard day, and beat a small group. Those are different riders and they are cheaper to hire.

So a stage-hunting squad selects for a different set of qualities: riders who read a race well enough to be in the right break, who can climb well enough to be in it on a hard day, who have a finish, and who can do it repeatedly rather than once. Consistency matters less. Variance is the point, because a team that needs one good day out of twenty-one should be maximising the chance of an outlier rather than the average.

The mountains classification is a related target and a genuinely rational one for a smaller team, since it is accumulated from breakaways rather than from being one of the fastest climbers in the race. A squad built to put a rider in the escape most days can chase that jersey while also being in position for stage wins, which is two objectives served by the same eight places.

The 23-team field makes this both more attractive and harder. More wild card teams means more squads whose season depends on three weeks of visibility, all of them attacking from kilometre zero on any day that looks winnable, which is why the fight to form the day's breakaway has become an event in itself. It also means more competition for the same escapes, so a team's stage-hunting plan needs riders who can go clear on days that look unpromising rather than only on the obvious ones.

None of this is a lesser ambition. It is a different reading of the same constraint, and it is usually the correct reading. The rest of our cycling coverage takes the same view of the sport's economics: the objective a team can afford is a better guide to its selection than the objective it would prefer.

What to look at when the next squad is announced

Four questions turn a list of eight names into a readable plan.

How many of them can climb late? Count the riders who could plausibly still be with the leader inside the final ten kilometres of a mountain stage. Three is a team riding for the overall. One is a team that has decided something else matters more, whatever the press release says.

Is there a sprinter, and did he bring anyone? A sprinter without a lead-out is a hedge rather than a plan, and often signals a team that could not decide. A sprinter with two dedicated helpers is a team that has committed a third of its squad to about six days.

Who is the oldest rider and what is he for? If the answer is not obvious from his results, he is probably the road captain, and his presence tells you the team expects the race to be decided by decisions rather than by watts.

And where is the slack? Find the rider whose job you cannot name. That is either the free role, deliberately unassigned so the team has an option when the plan fails, or it is a place that has been wasted. Which of the two it turns out to be is usually clear by the second rest day, and by then nobody can do anything about it.