Money, markets & the business of OCM
A market-value model for OCM riders, and why it is built the way it is: role, scarcity, age, and the state of the market.
The game gives every rider a rider value, and it is also the lowest price he can be sold for. Clubs nearly always pay more: 99.6% of transfers go above it, often at two to three times. To value a whole squad we need what a rider would actually fetch. Multiplying rider value by an average markup, the obvious shortcut, misses the typical S78 fee by 69%. This paper argues that a rider's price is set by four things: whether he can lead a team or only support one, how rare his skills are, how many good seasons he has left, and how much money active clubs have to spend. We show the evidence for each from five seasons of sales and set out a model that prices each of them separately. Our first rough versions cut the typical miss to 44%, though only just over half of sales land within ±50% of the estimate. Single sales stay hard to predict, so the model will give ranges, not single figures, and it will be judged on seasons it has never seen.
Rider value is the game's own number. It works like the value on a company's books: calculated by a formula, the same for everyone, and stable. It is also a hard floor. A rider cannot be sold for less than his rider value, and a club that releases a rider gets exactly his rider value back in cash. The smallest rider value attached to any transfer has been exactly 29,000 in every season since S35.
The price a club pays is a different number. It is agreed between two managers, and it depends on who is selling, who is buying, and what else is on the market. In seasons 73–77, clubs paid on average 2.36 for every 1 of rider value. In S78 that fell to 2.09.
In plain terms: rider value is the minimum the game guarantees. The fee is what people really pay. We want to predict the fee, and in particular how far above the minimum it goes.
The Rich List needs this for every rider in the game, most of whom will never be sold. The first version used the simplest rule possible: rider value times the season's average markup, with one markup for young low-value riders and one for everyone else. Tested against real S78 sales, that rule's typical miss was 69%. Fewer than two sales in five landed within ±50% of the prediction.
No two riders are alike, just as no two houses are. Economists handle this with a hedonic price model. A house is a bundle of features (bedrooms, a garden, distance to the city), and each feature raises its price by a certain percentage. Look at enough sales and you can work out what one extra bedroom adds, even though nobody ever sells a bedroom on its own.
A rider is a bundle too: a set of skills, an age, a role he can play in a team. If we can work out what each part of the bundle adds to the price, we can price any rider, including the 99% who are not for sale. Football economists have priced players' transfer fees this way for thirty years, and the same patterns turn up there: quality, age and scarcity drive the fee.
So the job is to find out which features buyers pay for, and how much each one adds.
Five seasons of sales give us 1,402 transfers where we know the rider's skills and age at the moment he was sold. Five patterns stand out.
A captain is the rider a team races for; teammates work for him. We measure captain quality with each rider's best captain line, the sum of the three stats his speciality needs: sprint is SP + FR + TQ, climbing is CL + DH + HL, cobbles is CB + FR + TQ, hills is HL + SP + FR, and for a time trial we count TT three times. Buyers pay a great deal for that line. A 22–23-year-old with a captain line of 225 sold for an average of 1.12M. A rider of the same age with teammate-level skills sold for about 163k, roughly a seventh as much.
What matters is not just how good a rider is but how few riders are as good. We ranked every sold rider's best captain line against all 12,000–13,000 riders with recorded stats that season.
Prices rise sharply near the top. Riders ranked in the top 50 on their best line sell for around 400–525k; riders ranked 51–250 for about 225–280k; riders below 1,000th for about 125k. The steps are not perfectly even (the 51–100 group sold a little below the 101–250 group in these five seasons), but the pattern is clear. Economists call it the economics of superstars: when only a handful can do something at the very top level, and there are few substitutes for them, small differences in talent turn into large differences in price.
Two details matter for the model. First, the elite riders who do get sold are old: their average age is about 30. Young elite riders almost never reach the market (section 3.5 explains why). Second, the same scarcity effect shows up in support roles. We measure those with four teammate lines: the sprint lead-out is SP plus half of TQ, the flat lead-out is FR plus half of TT, the cobble pacer is CB plus half of FR, and the climbing domestique is CL plus half of HL. Riders in the top 50 on one of those lines also sell in the 400k-plus band.
This is why we rank riders rather than just adding up their stats: a stat line is only worth a fortune if almost nobody else has it.
A rider is worth what he can still win, discounted for the seasons he has left. A good rider aged 24 has years of top form ahead; the same rider at 31 has a season or two.
For captain-grade riders the price falls steeply with age: a 210-line captain fetched about 460k at 22–23 and about 207k at 30 and over. For teammate-grade riders the line is almost flat at about 140–200k at every age. Buyers pay for a captain's future, but they barely pay for a teammate's.
Below the ranked stars sits a separate group: the youngest riders, ranked below 1,000th today, sell at 2.7 times rider value. They are not good yet. Buyers are paying for what they might become.
Prices are not fixed over time. S78 was a cool season: 240 deals against an average of 282, and 52.5M spent against 69.4M. The obvious explanation, that clubs had less money, is wrong: the cash clubs held grew by 3–5% over the season while prices fell. What sets the price level is how much money active clubs are willing to spend, and how many of them are buying. A large share of the game's cash sits with clubs that no longer race or trade, and that money does not bid for anyone.
OCM has no wages, so keeping a rider costs a club nothing. A club with a young star has no reason to sell him, and so the market mostly sees riders their clubs can spare: veterans, surplus teammates and prospects. That shapes every price in this paper. It is why the elite riders who do sell are about 30. It is also why teammate prices are flat: the hire list and releases keep a steady supply of serviceable teammates, so a club rarely needs to pay much for one. Filler riders sell at rider value plus a few thousand; riders with a talent buyers can see attract bidding wars that add 100k to 500k.
Putting those findings together, the model predicts how far above his rider value a rider would sell, on a log scale:
Each term is there for a reason we can point to in the data:
The skill lines will use the game's own race weights, so that "ranked 5th for sprints" means what the game means by sprinting, rather than the rough lines used in the tests above.
Two adjustments turn the model's answer into a value in the bank. First, converting a log prediction straight back into money gives a typical price, which runs below the average price, most of all for elite riders whose prices vary most. Summed over a squad, that would understate it, so the conversion is corrected for each kind of rider. Second, every value comes with a range: an 80% range for each rider and for each club, built so that a whole-market shock (a hot or a cold season) moves every rider together rather than cancelling out.
Clubs sell riders they can spare and keep the ones they cannot replace (section 3.5). The sales we learn from are therefore not a fair sample of all riders. Economists call this selection bias. The textbook correction needs something that changes whether a rider is sold without changing his price, and this game offers nothing like that. So we do not pretend to correct it. Instead, values for the riders the market rarely sees, above all young elite riders, get wider ranges, and the published pages say so.
Only about 50 riders ranked in the top 50 have been sold in five seasons. A handful of unusual sales can move the estimates. Using smooth curves instead of separate boxes is how the model copes, and its errors at the top will be reported separately.
About 280 riders change clubs in a season, out of roughly 12,000 on rosters. If one club tried to sell a whole squad at once, it would flood the market and prices would fall. So a squad's market value is not cash it could raise tomorrow. The Rich List will show two figures: the market value, and a lower "could actually sell" value that is discounted for how much of a season's demand that squad would add.
Clubs that no longer race or trade still own riders. Those riders will not come to market, and their owners' cash does not bid for anyone. Their squads will be flagged, so a reader can see how much of the list's wealth is active and how much is idle.
Even a good model will miss single sales by a wide margin, because one keen buyer can double a price. Random misses shrink as a share of a squad's total, but misses that share a cause, like a cold season or a type of rider the model gets wrong, do not cancel. The early tests show this: the season totals got slightly worse as terms were added. That is why squad and club totals get their own error checks and their own ranges.
Once values are published, managers may start quoting them, and future fees may drift toward them. The model would then be learning from its own echo. We will track how far fees sit from the estimates before and after publication, and treat repeat deals between the same two clubs, and extreme outliers, with caution. Values for individual riders will not be published until the model has a season of results behind it.
The test is strict: fit the model on earlier seasons, then predict a later season it has never seen, and repeat for several seasons in turn. The early tests below used one held-out season.
| Model | Typical miss | Within ±50% | Season total vs actual |
|---|---|---|---|
| Rider value × average markup (used today) | 69% | 39% | +12% |
| Add age | 45% | 54% | +14% |
| Add age and captain grade | 44% | 54% | +14% |
| Add age, captain and elite-teammate grade | 44% | 55% | +15% |
| Full model (this paper), target | below 44% | above 55% | within ±10% |
The full model has to pass all of these:
Once it passes the test, the same model serves several jobs:
It is not a price list and it is not the game's value. It says what a rider would be expected to fetch in a normal sale this season, not what any particular manager will pay. It cannot see a manager's plans, his cash, or the friend who will overpay. It is an estimate with a stated margin of error, and it gets better each season as more sales come in.
Reviewers pointed to things that matter to a manager but that this first model leaves out, mostly because the data to measure them is thin or missing:
Draft v2 (15 Oct 2026 · Season 79, Day 6). Revised after review by an economist, a game designer and an online-game economist. Changes: the model now predicts the markup over rider value and treats rider value as the game's minimum price; the price level is built from active clubs' money rather than one number per season; values are corrected when converted from logs and come with ranges; squads get a "could actually sell" value; dormant clubs are flagged; section 3.5 explains why stars rarely sell; the validation plan is tougher; the wording on adding up prices, misses cancelling and the "top 50 step" is corrected; skill lines are written out in full; references are corrected and football transfer-fee studies added.
Data: OCM club-to-club transfers S74–78, rider stats and ages as recorded at the time of each sale, ranked against every rider with stats that season. Hire-list signings carry no price and are not included. Figures in sections 3 and 6 come from preliminary tests; the full model's results will replace them when it is built.
The Echelon front page · The Rich List · The transfer market, seasons 73–77