Futarchy Labs · research on decision markets
Markets that answer “what if?”
Most markets tell you what something is worth. We study markets that tell you what it would be worth under each choice a group could make. We build them, run them with real money, and publish the code.
What we research
Markets as instruments for reasoning: what a group believes, how its beliefs fit together, and what its choices are worth.
Bayesian-network markets
Every question is a variable in one network. Move a probability, or a conditional one, and an inference engine keeps all the prices consistent.
Open Bayes Market
Decision markets
One asset priced once per choice. The gap between the prices is the market's estimate of what the decision is worth.
See the markets
Markets for AI agents
Teams of autonomous agents whose choices are made by internal markets instead of a fixed chain of command.
Agents
Mechanism design
How much money a market needs behind it, who should supply it, and what it costs to push a price the wrong way. Tested in simulation first.
SimulationsHow the markets work
A price is a forecast that someone paid to make.
We design markets where those forecasts answer the questions a decision depends on.
Bayes Market links questions in one network
Traders can say "if A happens, B is 70% likely", not only "B is 40% likely". A belief about one question updates the questions it bears on, the way it would in a careful forecaster's head.
A decision market prices one stock under each choice
Each trade only counts in the world it was made for and is undone otherwise, so every price is an honest answer to "what if this is chosen?".
People trade instead of voting
Anyone who thinks a price is wrong can correct it and is paid if they were right. Knowledge moves the price whoever holds it.
Arbitrage keeps the prices honest
Today's price has to equal the chance-weighted average of the prices under each choice. When it does not, correcting it is profitable.
Field experiments
Each of these ran with real money, and each asked the same kind of question about a different decision.
- What is NVIDIA stock worth under each possible next CEO?NVIDIAOpen
- What is NVIDIA stock worth if Jensen Huang stops being CEO?NVIDIAResolving
- What is an open-source protocol worth if a proposal to change it passes?Ethereum, Gnosis, KlerosClosed
Code
- bayes-marketBayesian prediction market engine, event-sourced and combinatorial
- simulationsMechanism-design simulations for decision markets
- agentsAgent teams governed by markets
- fire-the-ceoRobin Hanson's fire-the-CEO decision markets
- futarchyConditional markets with two or more options
Talks and writing
- Launching FutarchyDappCon 2025, Kelvin Santos and Robin Hanson
- Does AI governance need a Federalist Papers debate?LessWrong, 2023
- Price-Based FutarchyGovernance Day 2025
Guided by the Inventor of Futarchy
Robin Hanson - Chief Scientific Officer
Robin Hanson created the concept of futarchy in 2000 - using prediction markets for governance. As a professor at George Mason University and research associate at Oxford's Future of Humanity Institute, he developed the foundational theory of how markets can make better decisions than traditional voting or expert opinions. His paper "Shall We Vote on Values, But Bet on Beliefs?" shaped the field.
