Lmsr

What is the lmsr market maker?

An LMSR market maker (short for “logarithmic market scoring rule” market maker) is an automated market-making mechanism used in prediction markets to provide continuous buy and sell prices for outcome shares, even when there aren’t enough human traders to match orders. Instead of relying on a traditional order book, the LMSR sets prices based on how many shares of each outcome have been bought so far, so the price moves smoothly as traders express beliefs. This makes it a common tool for keeping prediction market liquidity available from the first trade to the last.

A practical way to understand LMSR is to focus on “how much it costs to move the price.” Suppose a binary market has outcomes YES and NO. If the current YES price is 0.60, buying YES shares pushes that price upward, but each additional share becomes slightly more expensive than the last because the cost function is convex. That convexity is the core design feature: it prevents the market from offering infinite liquidity at a fixed price, while still guaranteeing that anyone can trade immediately. An analogy: think of a thermostat with resistance—turning it a little is easy, but pushing it far from the current setting takes increasingly more effort. In LMSR, “effort” is the extra cost required to move the market’s implied belief.

Logarithmic market scoring rule

The logarithmic market scoring rule is the mathematical rule that links the market’s current state (how many shares of each outcome are outstanding) to the total cost paid by traders so far. In the standard formulation, the market maker maintains a “cost function” that grows as traders buy more shares, and the instantaneous price for an outcome is the marginal cost of buying one more share of that outcome. The “logarithmic” part comes from using a log-sum-exp style function, which ensures prices stay between 0 and 1 and that the market maker’s worst-case loss is bounded. A single parameter—often called b—controls how quickly prices move when someone trades.

Lmsr automated market maker

As an automated market maker, LMSR is always willing to trade at the current algorithmic price, so a participant doesn’t need to wait for someone else to take the other side. Conceptually, traders are buying “shares” that pay 1 unit if an outcome happens and 0 otherwise (for multi-outcome markets, exactly one outcome settles to 1). When a trader buys shares of an outcome, the outstanding quantity for that outcome increases, the cost function increases, and the implied probability-like price for that outcome rises; selling does the reverse. The liquidity parameter b acts like a depth knob: higher b means trades move the price less (better prediction market liquidity, but higher potential subsidy/loss for the market maker), while lower b makes prices jump more (cheaper for the market maker, but thinner trading).

Lmsr vs clob in prediction markets

LMSR and a clob (central limit order book) solve the same problem—price discovery—but in different ways. A clob aggregates bids and asks from traders and matches them when prices overlap; if few participants post orders, spreads widen and trading can stall, which hurts prediction market liquidity. LMSR replaces that dependency on organic order flow with an always-on quoting function, so there is always a tradable price, even in thin markets or long-tail questions. The trade-off is that LMSR typically requires a subsidy or risk budget (captured by the bounded-loss property and the b parameter), whereas a clob can operate with minimal operator risk but may fail to produce informative prices when participation is low.

Why the lmsr market maker matters

The LMSR market maker matters because it makes prediction markets usable in settings where participation is sporadic, outcomes are niche, or early trading would otherwise be impossible. By guaranteeing continuous pricing and bounding the market maker’s worst-case loss, LMSR provides a predictable way to “bootstrap” information aggregation without needing a deep pool of limit orders from day one. This is especially valuable when the goal is not just trading volume, but reliable, continuously updated forecasts. In that sense, LMSR is a foundational mechanism for building robust prediction markets that can surface collective beliefs even under low activity.