The Composed Market: the price of a market that did not exist until you asked for it
A market that exists has a price anyone can take. A market the customer composes has no price until someone asks for one: the legs are chosen, a model decides how likely they are together, and the number that comes back is deliberately not the product of the legs. This desk is about that number - where it comes from, why it is always shorter, what the operator refuses, and what happens when one leg of the combination never happens at all.
- samples
- 10 invented
- legs
- 3
- product of legs
- 5.60
- composed price
- 4.80
The two bars are drawn against one axis and the composed bar is 14.3% shorter. That single difference is the desk: 5.60 is what multiplication returns and 4.80 is what the legs actually have together, and the 0.80 between them is the whole of the operator's loss if it ever quoted the longer figure.
three legs, one match, three leg prices and one joint price; the gap is not a fee.
five endings, one queue: 52.8% of requests never reach a customer as a price they can take.
A composed market - a same-event combination or a market a customer asks for - is priced from how likely its legs are together, not from the product of their individual prices. On the samples three legs priced 1.60, 1.75 and 2.00 multiply to 5.60 but the combination is offered at 4.80, a 14.3% difference, because the legs are correlated: they happen together more often than independence implies.
What the samples show
The whole subject sits in one subtraction. Three legs in one match price at 1.60, 1.75 and 2.00; multiplied they give 5.60, which is the answer every reader would compute. The combination is offered at 4.80. The 0.80 of price between the two figures is not a fee and not a margin adjustment: it is the difference between a joint chance of 17.86%, which is what independence would imply, and the 20.83% the model measures.
That distinction matters commercially. Priced at 5.60 against a real 20.83% chance the leg-set returns 116.65 per 100.00 staked, so the book loses 16.6% of every stake placed on it. The composed price of 4.80 is not greed; it is the price at which the bet stops being a gift. The same arithmetic explains the second finding: what the customer asks for is refused more often than it is granted - 218 of 500 requests in the sample month were accepted, and 131 never became a market at all.
None of the samples describes a real operator, sport, player or match. They are ten invented sets of counts and prices, defined on this page, and every other figure on the site is derived from them.
Ten samples
Three legs, one match.
- product / price
- 5.60 / 4.80
- difference
- 14.3%
- joint chance
- 20.83%
One month of requests for a market.
- requests
- 500
- accepted
- 218
- no market
- 131
Forty documented rules.
- permitted
- 22
- manual price
- 11
- refused
- 7
A leg that never happens.
- four legs
- 13.44
- offered
- 10.80
- repair error
- 6.7%
How far a combination will go.
- legs
- 2 to 12
- ceiling
- 500.0
- states at 12
- 4,096
Forty terms pages, read.
- state a model prices it
- 26
- state it is not the product
- 4
- all six rules
- 1
One month of combination bets.
- turnover
- 170,400.00
- priced independently
- 16.6%
- a year
- 339,436.80
What answering requests costs.
- a request
- 41.07
- a month
- 20,535.00
- wasted
- 5,380.17
Two further samples are defined on the pages that use them: sample I on the reasons a request is refused, and sample J on how wide the model's answer really is.
The desk in one table
The clearest place to start is the figure the whole subject turns on: what the customer computes, and what the operator returns.
| Step | Price | Chance it implies | Where it comes from |
|---|---|---|---|
| The three legs, multiplied | 5.60 | 17.86% | Each leg's own price, treated as independent |
| The combination as offered | 4.80 | 20.83% | The model's measured joint chance of the three together |
| What the difference is worth on a 10.00 stake | 0.80 | 2.97 points | 56.00 against 48.00 returned |