Market-Based Forecasting Engine
Aggregate informed financial stakes into a live probability forecast
- Difficulty
- Advanced
- Time to result
- ~ongoing to results
- Steps
- 6
- Confidence
- 94%
Define an event as a binary contract whose final outcome can be settled using an authoritative source. Participants who think the event is more likely buy one side, while participants with the opposite view take the other. Because money is at risk, traders have an incentive to reveal sincere beliefs, research the event, and act when their information differs from the current price. The matching process aggregates these distributed judgments into a continuously changing market price, which can be read as an implied probability. The output becomes a live forecast that incorporates new information faster than periodic polling or commentary. Its reliability, however, depends on clear rules, broad participation, and enough liquidity that isolated orders cannot dominate the signal.
Origin
Extracted from Masters in Business, where Luana Lopes Lara explained Kalshi's long-term ambition to make event-market prices a widely used source of forecasting data.
Core principles
- 01Require participants to bear a cost for inaccurate beliefs
- 02Aggregate many independent information sources through trading
- 03Express prices as interpretable probabilities
- 04Use objective settlement rules
- 05Demand sufficient liquidity before trusting the signal
How to run it
- 1
Specify the event
Frame one future outcome as a clear yes-or-no question with a defined deadline.
Pro tip Test whether two neutral readers would interpret the question identically.
Watch out Vague concepts create disputes and contaminate the forecast.
- 2
Define settlement
Name the authoritative data source, threshold, timing, and edge-case rules that determine the result.
Pro tip Write the rules as if they were a legally binding contract.
Watch out Do not rely on discretionary judgment after trading begins.
- 3
Attract opposing information
Bring together participants with different exposures, models, knowledge, and beliefs about the event.
Pro tip Seek both natural hedgers and analytical speculators because they contribute different information.
Watch out A homogeneous participant pool can amplify shared biases.
- 4
Let stakes reveal conviction
Allow participants to buy yes or no positions at prices reflecting their confidence and expected value.
Pro tip Watch actual trades rather than unactioned opinions.
Watch out Financial stakes improve incentives but do not eliminate manipulation or error.
- 5
Read the implied probability
Interpret the contract's cent price as the market's current probability estimate and track how it changes after new information.
Pro tip Pair the probability with volume and spread data to judge signal quality.
Watch out Do not present a thin-market price as precise certainty.
- 6
Use and recalibrate
Incorporate the forecast into planning, hedging, or communication, then compare final outcomes with prior probabilities to assess calibration.
Pro tip Evaluate many forecasts over time rather than judging the method from one result.
Watch out A well-calibrated forecast can still be wrong on any individual event.
In the wild
A contract asking whether a recession will occur during the year trades at twenty cents. Its rules specify the authoritative determination and settlement date. Participants continuously trade as employment, inflation, and output data arrive, turning the price into an updated estimate of recession probability.
→ Decision-makers receive a live, financially informed probability rather than waiting for occasional surveys or pundit forecasts.
A company creates an internal binary market on whether a major release will ship by quarter-end. Engineers, salespeople, and operations staff trade using private credits, while the settlement rule points to the production deployment record.
→ Leadership gains an aggregate schedule forecast that incorporates information dispersed across teams.
Common mistakes
Using ambiguous outcomes
If settlement depends on interpretation, the market price reflects rule uncertainty as well as event probability.
Ignoring liquidity
Thin markets can move sharply on one order and may not represent a stable aggregate belief.
Treating probability as certainty
A quoted probability represents uncertainty and should inform decisions rather than guarantee an outcome.
Is it for you?
Best for
It is best for clearly defined uncertain events with diverse informed participants and an objective settlement source.
Not ideal for
It is not ideal for subjective outcomes, ethically unacceptable markets, or thin markets where one trade can distort the price.
From the transcript
“because when people put money where their mouth is, they are more more likely to say what they really think and actually do research and…”
“markets really aggregate the wisdom of the crowds.”
“I think we need a base level of liquidity and volume for for the forecast to really work and be really useful.”
From the episode
Luana Lopes Lara on Kalshi’s CFTC Approval Journey
Luana Lopes Lara