What Is LMSR Pricing in Prediction Markets?
LMSR (Logarithmic Market Scoring Rule) is the most widely used automated market maker for prediction markets — it provides continuous liquidity and mathematically bounded risk, making it the backbone of platforms like Polymarket and Kalshi.
LMSR (Logarithmic Market Scoring Rule) is an automated market maker algorithm designed specifically for prediction markets. Created by economist Robin Hanson, it allows a prediction market to offer continuous trading without requiring a counterparty for every trade — the algorithm itself acts as the market maker, adjusting prices based on demand.
The Core Problem LMSR Solves
Traditional order book exchanges (like stock markets) need buyers and sellers to match. In prediction markets — especially niche or long-tail events — there often aren't enough participants for organic price discovery. You might have someone who wants to bet $50 that a specific policy will pass, but no one on the other side at that moment.
LMSR solves this by guaranteeing liquidity: any trader can buy or sell at any time, with the algorithm dynamically adjusting the price based on accumulated demand.
How It Works (Simplified)
LMSR uses a cost function to determine prices. For a binary market (Yes/No):
- The market starts at 50/50 (or whatever the operator sets)
- Each purchase of "Yes" shares pushes the Yes price up and the No price down
- Each purchase of "No" shares pushes the No price up and the Yes price down
- The price changes are logarithmic — small trades have small impact, large trades have larger impact
The key parameter is b (liquidity parameter):
- Low b = prices move quickly with small trades (thin market, more volatile)
- High b = prices move slowly, requiring larger trades to shift (deeper market, more stable, but higher operator subsidy)
The operator's maximum loss is mathematically bounded at b × ln(n) where n is the number of outcomes. This is the "subsidy" the operator pays to guarantee liquidity — essentially the cost of running the market.
Why Operators Choose LMSR
| Advantage | Description |
|---|---|
| Guaranteed liquidity | Every market is tradeable from launch — no cold-start problem |
| Bounded risk | Maximum operator loss is known upfront before the market opens |
| Price discovery | Prices naturally converge toward true probabilities as traders participate |
| Simplicity | No need to manage order books, match trades, or recruit market makers |
| Composability | Works for binary, categorical, and scalar markets with minor modifications |
LMSR vs. Other Approaches
Order Books (traditional exchanges): Better for high-volume markets with many participants. Poor for long-tail prediction markets where liquidity is sparse.
CPMM (Constant Product Market Maker): Used by Uniswap and similar DeFi protocols. Simpler math but less capital-efficient for prediction markets. Higher slippage on large trades.
Dynamic Parimutuel: Used by some sports betting platforms. No market maker risk, but prices only finalize at market close — no continuous price discovery.
LMSR hits the sweet spot for prediction markets: continuous pricing, bounded risk, and reliable liquidity even for markets with few participants.
Real-World Usage
- Polymarket uses a modified LMSR (CLOB + AMM hybrid) for its largest markets
- Kalshi uses LMSR-style pricing for many event contracts
- Metaculus and Manifold Markets use LMSR variants for their community prediction platforms
- Corporate prediction markets (Google, formerly at Microsoft) used LMSR for internal forecasting
What Operators Need to Know
When implementing LMSR for a prediction market product:
- Set b carefully — Too low and prices are noisy; too high and you're subsidizing heavily
- The subsidy is your cost of liquidity — Think of it as a marketing expense for running the market
- Multi-outcome markets need more subsidy — b × ln(n) grows with outcome count
- Consider dynamic b — Some implementations adjust liquidity based on trading volume
Adkuu Pulse takes the fixed-odds route instead of an LMSR market maker: it delivers ready-to-use fixed odds — an AI probability shaped by a built-in risk model, plus your margin — so operators offer prediction markets as classical fixed-odds betting. The raw probabilities are still available if you'd rather run a peer-to-peer market.
Last verified: March 2026