One market per number, priced algorithmically. The market
maker funds it, sets its opening view, and can manage its risk while it
trades.
Pricing numerical outcomes
functionSPACE aims to be the standard for pricing numerical outcomes.
Most questions worth trading have a number for an answer: a rate decision, an
inflation print, a closing price, a match score. Binary venues split each
number into a ladder of yes/no markets, each with its own liquidity. A
functionSPACE market quotes the whole distribution over the number, and
traders buy payoffs on a value, a range, or any shape of outcomes against one
pool of capital.
Why an automated market maker
An order book would need resting quotes on every bucket of that
distribution, and its liquidity would fragment across them. An automated
market maker (AMM) prices every bucket algorithmically from the same pool,
while keeping the risk for the market maker bounded.
Prediction-market AMMs have nonetheless lost ground to order books. We
believe the reason is that market making involves risk, and market makers on
prediction-market AMMs have not been given the actions to manage it. Our mechanism is
built around those actions.
The market maker’s position
The market maker (MM) is the party that funds the AMM and carries its risk, and
that risk is bounded: the MM can lose at most the capital it has in the
market. The MM opens the market with a prior, its own distribution
over the outcome, much as a bookmaker sets opening odds; the crowd then
calibrates it. A floor on every prior keeps any outcome from being quoted
away. The MM earns a flat commission on every trade, fixed by the protocol
and charged on top of the price. By default the commission is set aside
where no later trade can reach it; the MM may instead reinvest part of it as
depth.
The MM’s prior sets the opening odds; trading moves the market
from there. A trader’s position on a range
is priced off the market’s distribution.
Computed with the mechanism’s own pricing code.
functionspace.dev1 / 2
MECHANISM v0.5.0 · SEPTEMBER 2026
Managing risk on a live market
Bookmakers and exchange market makers can move their line and pull their
quote. Prediction-market AMMs could not: the price moved only with trades,
and the only way out was to leave the market, so the MM stayed committed to
its opening view whatever new information arrived. Version 0.5.0 of our
mechanism gives the MM four actions on a live market.
Add and remove liquidityScale exposure up in a market that trades well; take capital out
of one whose risk no longer pays.
Reshape the priorMove the line on new information, without erasing the positions
already taken.
Derisk against incoming volumeKeep part of each incoming buy out of the pool, selling exposure
down as the flow arrives.
Withdraw earned commissionAt any time. Commission set aside is never at risk in the market.
None of these actions changes a trader’s settlement claim.
Left, the market from page 1. News arrives and the MM reshapes
its prior (right): the market’s distribution moves with it, without a
trade. Positions keep their settlement payout; their price before
settlement moves with the market. Computed with the mechanism’s own
pricing code.
Interactive notebooks
See how the derisking works
The same notebooks price each of the four actions with the mechanism’s own
code, in your browser, with your parameters: what adding or removing
liquidity does to the market maker’s exposure, what moving the prior costs,
how routing part of each fill sells exposure down as the flow arrives, and
what the commission earns. The abridged technical writeup comes with
access.
Interested parties can get in touch to receive an abridged technical
writeup of the full mechanism and access to our interactive notebooks. The
notebooks run the mechanism’s pricing code in the browser, so you can
explore the quantitative questions with your own parameters: the MM’s
returns, how profit and loss depends on the prior and on order flow, what
each action costs, and which markets suit the maker best.
