LS-LMSR: How Liquidity-Sensitive Pricing Works in Prediction Markets
LS-LMSR is a technical term, but the problem behind it is practical. How can a prediction market offer a usable price before many traders arrive? In a new market, a small order can move the implied probability too much. In a mature market, a price curve that is too rigid can hide real demand.
LS-LMSR tries to balance those two states. It adapts market-maker liquidity as the market grows, instead of treating every market as if it had the same depth. That matters for anyone who wants to understand probability, slippage, execution, and prediction-market microstructure.
This guide explains what this mechanism is, how it differs from standard LMSR, how it compares with order books, and how traders can read liquidity before entering a position. It is educational only, not trading advice.
Last reviewed: May 2026.
What Is LS-LMSR?
LS-LMSR means Liquidity-Sensitive Logarithmic Market Scoring Rule. It is a variant of LMSR designed for automated market making in prediction markets. The key idea is simple: the market maker does not use one fixed level of liquidity forever. It can adjust depth as the state of the market changes.
In plain English, LS-LMSR is an adaptive AMM for prediction markets. It does not wait for a buyer and seller to meet in an order book. Instead, a cost function quotes prices and updates implied probabilities after each trade.
The important word is “liquidity-sensitive.” In classic LMSR, liquidity is governed by a fixed parameter. In LS-LMSR, liquidity can respond to market size or activity. The price curve can become less fragile when there is more trading interest.
The model is discussed in academic work on automated market makers, especially A Practical Liquidity-Sensitive Automated Market Maker by Othman, Pennock, Reeves, and Sandholm. It also appears in broader comparisons of AMM models such as Automated Market Makers for Decentralized Finance.
The Problem LS-LMSR Solves
Every prediction market has to answer one basic question: who provides the first tradable price? In a central limit order book, traders must post bids and offers. Without those orders, the market can look empty.
An AMM solves that by using a rule. The rule always offers a price. That is useful for new markets, niche topics, and long-tail events where organic liquidity may not exist yet.
But there is a trade-off. If the curve is too shallow, a small trade moves the price too far. If the curve is too deep, the platform may need to subsidize more liquidity than the market can justify. LS-LMSR exists to make that trade-off more dynamic.
New-market liquidity
Imagine a market on the next central bank decision. On day one, only a few traders have arrived. Someone still wants to buy “YES” shares.
Without liquidity, that trader may not find a counterparty. With an AMM, the trade can happen against the pricing rule. That improves the user experience, but the rule must be calibrated.
If the market is thin, a $100 order might move the implied probability from 50% to 60%. In a deeper market, the same order might move it from 50% to 51%. LS-LMSR tries to make the second behavior more likely when the market has enough activity.
Slippage and price impact
Slippage is the gap between the price a trader expects and the average price actually paid. In prediction markets, it appears when the order itself changes the probability while it is being filled.
In a shallow market, buying a lot of “YES” pushes the marginal price higher. The trader starts at a lower probability and ends at a higher one. The average entry price is worse than the first number on the screen.
LS-LMSR does not remove slippage. No serious market design can do that for free. It tries to make price impact more reasonable as the market becomes more active.
Platform risk
An automated market maker has to be funded. More liquidity means better execution, but it can also increase the economic exposure of the market operator.
Classic LMSR has an elegant feature: maximum loss can be bounded by the liquidity parameter. But choosing that parameter is hard. Set it too high and the subsidy can be expensive. Set it too low and the market feels unusable.
LS-LMSR adjusts that relationship. It can offer more depth when the market shows demand and less exposure when the market is still small.
How LS-LMSR Works in Practice
The intuition has three parts: a cost function, a marginal price, and outcome shares. In a binary market, the outcomes are usually “YES” and “NO.”
The cost function defines how much it costs to increase the number of shares for one outcome. The marginal price is the price of the next unit. If many traders buy “YES,” the implied probability of “YES” rises.
In LMSR, the slope of that curve is controlled by a fixed liquidity value. A larger value creates more depth. A smaller value creates more sensitivity.
In LS-LMSR, that depth can change. It may respond to market size, accumulated volume, outstanding shares, or another state variable chosen by the mechanism designer. Implementations can differ, but the goal is the same: adaptive liquidity.
A simple election example
Suppose a binary market asks: “Will Candidate A win the election?” The initial “YES” price is near 50%. That is an implied probability, not an official forecast.
In a shallow market, a $1,000 buy order may move “YES” from 50% to 58%. The order had large impact because the curve was thin.
In a deeper market, the same $1,000 order may move the price from 50% to 52%. The impact is lower because the curve has more depth.
LS-LMSR aims to move the market toward that second behavior when activity justifies it. At the same time, it avoids offering excessive depth before demand exists.
At Macro Markets, we use this kind of framework to separate a clean probability signal from a weak execution setup. The important habit is not to memorize formulas. The habit is to ask whether the displayed price can survive a real order.
LMSR vs LS-LMSR
The core difference is liquidity. LMSR uses a fixed liquidity parameter. LS-LMSR makes liquidity sensitive to the state of the market.
Think of the comparison this way:
- LMSR: fixed cost function, fixed liquidity, simpler to explain, but one parameter can be too shallow or too expensive.
- LS-LMSR: adaptive cost function, liquidity changes with market state, better for growing markets, but harder to implement and explain.
- CLOB: prices come from trader orders, liquidity depends on buyers and sellers, strong when markets are active, but weak when books are empty.
LS-LMSR is not automatically better in every case. It is a tool for a specific problem: providing algorithmic liquidity without treating small and large markets the same way.
LS-LMSR, CLOBs, and AMMs
A central limit order book forms prices from trader orders. Buyers post bids. Sellers post offers. The market price emerges from competition between them.
An AMM forms prices by rule. A trader does not need a direct counterparty at that moment. Execution can happen against the cost function.
Public Polymarket documentation describes a hybrid CLOB with off-chain matching and on-chain settlement. Kalshi’s public API exposes order-book endpoints for binary markets. These examples help compare market designs, but they do not imply those platforms use this mechanism.
Where order books are strong
Order books work well when many participants compete. Spreads can tighten. Prices can update quickly. The visible book also helps traders understand available depth.
The weakness appears in new or niche markets. If nobody posts orders, execution is limited. Interest in the event may exist, but tradable depth may not.
Where AMMs are strong
AMMs are strong at launch. They make a new market tradable immediately. That is useful for long-tail prediction markets.
The weakness is calibration. A curve can be too shallow, too deep, or poorly matched to risk. LS-LMSR matters because it makes the curve more dynamic.
Hybrid designs
Some market designs can combine both approaches. An AMM can support early liquidity while user orders become more important as the market grows. A platform can also use different mechanisms for different market types.
The best design depends on product goals, regulation, risk, and user behavior. High-volume markets may support a competitive order book. Smaller markets may need more automated liquidity.
What Traders Should Watch
The displayed probability is only one signal. The better question is: what probability do you actually pay after your full order is executed?
Before trading, check seven things:
- Depth: how much does the price move for your intended size?
- Spread: is there a large gap between buying and selling?
- Recent volume: is activity current or only historical?
- Market size: does the topic attract enough participants?
- Price impact: does your own order change the probability too much?
- Resolution rule: is the event outcome clear and verifiable?
- Exit liquidity: can you reduce the position before resolution?
This checklist applies to AMMs, LS-LMSR, CLOBs, and hybrid markets. The technology changes, but the trading question stays the same: what is the real cost of entering and exiting?
Volume is not liquidity
Volume measures what already traded. Liquidity measures what can be executed now. That difference is critical.
A market can have large historical volume and still be thin today. Another market can have low volume but solid algorithmic depth. LS-LMSR tries to make that second case more viable.
Simple rule of thumb
Price is a signal. Cost is the test. If the signal looks good, test the cost. If the cost is too high, reduce size, wait, or skip the trade.
If a small order moves the price from 50% to 51%, the market is relatively firm. If the same order moves it to 58%, execution risk is high. The event view may still be right, but the trade may be too expensive.
FAQ
Is LS-LMSR the same as LMSR?
No. LS-LMSR is derived from LMSR, but it adds liquidity sensitivity. Classic LMSR uses a fixed liquidity value. LS-LMSR can adjust depth as the market changes.
Does LS-LMSR eliminate slippage?
No. It can reduce relative price impact in more active markets, but it does not remove slippage. Large orders can still move the probability.
Is LS-LMSR better than an order book?
It depends. LS-LMSR can be better for early liquidity and continuous execution. An order book can be better when many traders compete with real bids and offers.
Why does this matter for prediction markets?
Prediction markets need both probability and execution. If the probability looks good but liquidity is weak, the price may move too much before the order is complete.
Conclusion
LS-LMSR is best understood as adaptive liquidity for prediction markets. It starts from LMSR, but it lets market depth respond to the state of the market.
The practical lesson is simple. New markets need initial execution. Mature markets need deeper liquidity. A fixed parameter can fail at both extremes.
For traders, do not look only at the displayed probability. Watch slippage, depth, spread, recent volume, and exit liquidity. In prediction markets, price and microstructure move together.
For platforms, this design shows that liquidity is not a detail. It is part of the product. A well-designed market must form useful probabilities and allow execution at a clear cost.
