Automated Market Maker for Prediction Markets
In Simple Terms, What Does the AMM Do?
A few simple steps, in order. No jargon needed.
The AMM injects initial YES and NO liquidity from an allocated budget, and continuously refills liquidity whenever the order book becomes empty, so the market never looks empty to a visitor.
Whenever a side's liquidity runs low or the order book empties out, the AMM refills it automatically, scaled to recent demand, so trading never stalls.
Real users always match each other first. The AMM matches only when real users are not available, then exits its position through a sell order.
It adjusts prices, manages exposure, and keeps reporting P/L and risk to your admin dashboard so operators stay in control.
What Is an AMM, and Why Prediction Markets Need One
A prediction market lets people trade on the outcome of real-world events: each market has a YES side and a NO side, and prices reflect the crowd's probability of the outcome. But a brand-new market starts empty. There are no orders to trade against, so the first visitors see a blank book and leave. This is the cold-start problem.
An Automated Market Maker (AMM) fixes that. It is software that seeds both sides of the market with YES and NO bets from an allocated liquidity budget, so there is always something to trade against. When a real user's bet matches an AMM bet, the AMM exits that position through a sell order. It never buys real users' sell orders. Our AMM is built specifically for order-book (CLOB) prediction markets, with admin-controlled budgets, intelligent pricing, and protections that keep real users safe.
Instant TradabilityDay 0
Liquidity is placed hours before the event goes live, so the very first visitor sees a populated order book with sensible prices on both sides.
Prices That Track RealityHybrid pricing
AMM bets don't sit still. A hybrid CLOB + LMSR model moves them as trading happens, blending live order-book signals with inventory-based risk pricing.
Capital Under ControlBudget caps
Every market gets an explicit budget, split per side if needed. The AMM can never deploy more than what the admin team has allocated.
Fully AuditableReports
Every action is logged. Daily P/L, ROI, risk, exposure and liquidity reports are generated automatically and exportable for commercial review.
Most automated market makers were built for crypto token swaps. Ours was engineered from the ground up for event markets and the operators who run them.
Designed specifically for YES/NO event markets, not adapted from crypto swap pools.
Operates inside real order-book markets, not only pool-based pricing models.
Independent budget control for every market and every side, enforced as hard caps.
Live order-book signals blended with inventory risk pricing for smarter, smoother movement.
Real users always match each other first. AMM liquidity steps in only when no real counterparty exists.
P/L, exposure, risk, liquidity, and audit reports designed for commercial and compliance teams.
Allocate budget & configure
Place initial YES/NO bets
Smart refill on demand
Hybrid price intelligence
Shift & protect the book
Sell at spread, track P/L
The Full Liquidity Lifecycle, End to End
From the moment a budget is allocated to the moment a position is sold, the AMM follows a controlled, fully logged sequence. Here is the journey of every unit of liquidity.
Example market configuration
Configure & Allocate Budget
The admin team enables the AMM per event and configures each market individually: total budget, YES/NO side budgets, starting probabilities, spread, grid depth and the LMSR liquidity factor.
- Liquidity mode: provide both sides, YES only, or NO only.
- Hard budget caps: budget_yes + budget_no ≤ budget, enforced server-side with row-level validation errors.
- Sanity rules: probabilities must sum to 100%, spread must be at least one tick, grid limited to 1 to 20 levels.
- Safe setup window: AMM can only be configured before any real order exists on the event.
Each bar is one price level below the anchor. The smallest amount sits at the anchor price, with larger amounts deeper in the book. A higher taper rate makes that increase steeper; a lower rate keeps the levels more even.
Initial Liquidity Placement
Hours before the event opens (a configurable window), the engine places the first AMM bets on the YES and NO sides. It deliberately deploys only a randomized slice of the budget, typically 3 to 8% per side, so the book never shows predictable, robotic patterns. The full budget remains the hard cap for everything that follows.
Bets are placed on a tapered grid: the smallest amount sits at the anchor price (derived from the configured probability), with progressively larger amounts at each tick below, so the deepest liquidity sits below the anchor. Placement is idempotent: if AMM bets already exist, it safely exits without doubling up.
Smart refill formula
Smart Refill: Liquidity That Follows Demand
When real users consume more than 80% of a side's AMM liquidity, the engine refills it, but intelligently. The refill amount scales with recent user volume and velocity over the last six hours, so busy markets get deeper books and quiet markets don't waste capital.
- Activity gate: no recent real-user volume means no refill, unless consumption is critical (above 95%).
- Budget recycling: realized sell proceeds extend the side budget until the final 3 hours before expiry, so working capital keeps working.
- Guardrails: refills are bounded between a randomized minimum and the remaining side budget. Never beyond the cap.
- Event-driven: refill and pricing jobs are also queued automatically after every trade, not just on schedule.
| Exposure | Risk band | Movement |
|---|---|---|
| ≤ 0.50 | Band 0 · Healthy | 100% of δ_max |
| ≤ 0.75 | Band 1 · Watch | 80% of δ_max |
| ≤ 1.00 | Band 2 · Caution | 60% of δ_max |
| ≤ 1.50 | Band 3 · Restricted | 40% of δ_max |
| > 1.50 | Band 4 · Defensive | 20% of δ_max |
Exposure-aware risk bands: the AMM automatically gets more conservative as risk grows
Hybrid CLOB + LMSR Pricing
Where should AMM bets sit as the market moves? The engine blends two signals: the live order book (what traders are actually paying) and an LMSR inventory price (what the AMM's own position implies about risk). The blend is damped, risk-banded and clamped so prices move smoothly, never lurching.
As the AMM's exposure grows, the risk band tightens the maximum price movement automatically, from full freedom at low exposure down to 20% of normal movement at very high exposure.
Order safety rules
Order Shifting & Real-User Protection
With a new reference price computed, the engine repositions the book: unfilled AMM bets shift toward the target, partially matched bets are cleanly cancelled and refunded, and sell orders are realigned to preserve the configured spread.
Critically, the AMM defers to real users:
- Match priority: real-user bets match against other real users first; AMM liquidity steps in only when no real-user counterparty is available.
- Never crossing real users: an order-book verification pass moves or cancels any AMM order that would cross a real user's order or compete too aggressively with it.
- Self-match guard: a lobby-side check prevents users from accidentally placing bets that could only match their own orders or AMM-only liquidity.
Exit order mechanics
Sell Orders & P/L Capture
The moment a real user's bet matches against an AMM bet, the engine immediately places a sell order at the configured spread, with a sensible floor. This is how the AMM exits inventory and creates spread opportunities. The AMM never buys: it only places initial liquidity bets and exits positions through sell orders. It never buys real users' sell orders.
- Spread capture: once a real user matches an AMM bet, the position is offered for sale at matched price + spread.
- Settlement outcomes: positions held to expiry are settled at face value and recorded in P/L, whether favorable or not.
- Price-capped: sell prices are always clamped within the market's valid range.
Liquidity provision can become a measurable commercial engine when spread capture, inventory, risk, and settlement outcomes are tracked properly. Every match is followed through to realized P/L, and ROI is computed daily per event, per market.
AMM bets matched by real users are offered for sale at the configured spread above the matched price. Each completed round trip is recorded as profit_sell.
Inventory held to expiry is settled at face value when the event resolves. Favorable settlements are recorded as profit_win; unfavorable ones flow into max-loss tracking.
P/L is always measured against what was actually deployed, so commercial teams see true capital efficiency per market.
A commercial-grade liquidity product, built in and battle-ready, not a roadmap.
Independent budgets, probabilities, spreads, grids and liquidity modes for every market in every event.
Every configuration row is validated on the server with precise, row-level error messages, even if client checks are bypassed.
Quote both sides, YES only, or NO only, with independent side budgets enforced as hard caps.
1 to 20 price levels with a tunable taper rate controlling how liquidity concentrates near the anchor price.
Initial placements use randomized sizing to avoid robotic patterns, and safely no-op if liquidity already exists.
Refills scale with real user volume and velocity, recycle sell proceeds, and skip dead markets automatically.
Order-book signals blended with inventory risk pricing, with damping, deviation clamps and 5 exposure-based risk bands.
AMM bets and sell orders track the reference price on both sides while always preserving the configured minimum spread.
Real users always match each other first; AMM liquidity matches only when no real-user counterparty exists, and AMM orders never cross real users.
Operations can pause or cancel AMM liquidity per event or per market at any time; cancellation withdraws open AMM bets.
Every admin action, including budget changes, grid changes and status changes, is written to liquidity logs with before/after snapshots.
A complete admin API surface: dashboards, P/L audit, exposure audit, risk views, liquidity views, and CSV exports.
The AMM is an operator liquidity tool, not a profit machine. It runs under explicit budget controls, full logging, exposure reporting and admin oversight, so it protects your balance sheet while it protects your markets.
The admin dashboard gives operators the power to set hard stops and cap maximum liability per market. Injection, refill and recycling all respect the configured per-side limits; the AMM cannot exceed its allocation.
Pause liquidity placement instantly, per event or per market, from the admin dashboard. Cancellation withdraws open AMM bets immediately. A global setting gates every automated job.
As cost approaches the LMSR loss threshold, risk bands automatically shrink allowed price movement, down to 20% in defensive mode.
Every 30 minutes, a risk report computes position imbalance, a VaR-style risk figure and max-loss per event, finalized at settlement.
Every admin action and every liquidity event is logged with before/after snapshots, giving compliance teams a defensible record of operations.
Nothing runs without operator-approved configuration. Budgets, spreads, grids and risk parameters are set by your team, and visible at all times.
The AMM engine ships with a full admin console: setup, liquidity monitoring, P/L dashboards, exposure audit and risk views. Here is what your operations team works with.
Market setup with per-side budgets, spread and grid controls
Live liquidity monitoring: injected, open, matched, cancelled
Risk dashboard with exposure, max loss and VaR-style metrics
Continuously Improved, Version by Version
v2.0.0Major Release · 27 May 2026
v2.0.0Major Release · 27 May 2026
Administration, dynamic liquidity, pricing controls & risk
AMM administration & configuration: centralized configuration management from Admin, controls to pause, edit and cancel AMM operations, and system user management for AMM-generated activity and order handling.
Liquidity management: dynamic increase and decrease of AMM budget from Admin, outcome-wise liquidity injection controls (Yes, No, or both outcomes), and advanced AMM reallocation for optimized liquidity distribution.
Pricing & spread controls: configurable spread management, price aggression management for controlled market movement, grid management for liquidity layering and order placement, and resolution of zero-spread issues for pricing stability.
Risk & validation: conflict order validation for buy and sell orders, plus a dedicated risk dashboard for exposure and operational monitoring.
Reporting & monitoring: a comprehensive AMM reporting suite including dashboard reports, revenue reports and exposure reports, with improved visibility into performance, liquidity behavior and operational metrics.
v1.0.0Initial Release · 29 December 2025
v1.0.0Initial Release · 29 December 2025
Core AMM engine foundation
Global AMM configuration: centralized global configuration for AMM behavior, configurable parameters for liquidity limits, pricing thresholds and exposure control, and the ability to enable or disable AMM participation per environment.
Market-wise budget & exposure control: market-level budget allocation, controlled capital deployment per market to limit downside risk, and automatic prevention of AMM over-exposure in volatile markets.
AMM reporting & monitoring: dedicated reports for matched volume, exposure and P/L, market-wise and time-based performance visibility, and operational transparency for audit and compliance review.
Initial liquidity provisioning: automated initial liquidity seeding for newly created markets, ensuring early price discovery and smooth user onboarding.
AMM & user order matching: AMM orders participate directly against real user buy and sell orders, ensuring continuous market availability when user liquidity is low.
Post-match AMM sell order placement: after a successful AMM-to-user match, the AMM places a corresponding sell order at the matched price to maintain market depth and balanced order books.
Anti self-trade protection: built-in protection preventing AMM orders from matching other AMM-originated orders, avoiding self-trading and ensuring every AMM trade represents genuine interaction with real users.
Four report families are generated automatically: daily P/L at midnight, risk and liquidity snapshots every 30 minutes, all served through a full set of dashboard and export APIs.
Per event, market and date: deployed budget, matches, cancellations, spread P/L, settlement P/L, and ROI.
Allocated vs. used budget, YES/NO positions, P/L, max loss, and a VaR-style risk figure built on position imbalance.
Injected, open, matched and cancelled liquidity per event, with market and category filters for the dashboard.
An immutable trail of every create, enable, pause, cancel, budget and grid change. Who did what, and when.
Launching new markets with limited initial liquidity and a cold-start problem to solve.
Teams that want order-book style markets but need automated liquidity support behind them.
Adding opinion and event markets alongside existing gaming products.
Adding event-based trading to an existing product with institutional-grade controls.
Anyone who needs exposure control, risk reports, audit trails and hard budget limits before they switch automated liquidity on.
Stories of challenges solved and value created.
Explore sports markets, leagues, and real-time event coverage.
Trade on crypto events, trends, and market movements.
Integrate with real-time markets, data, and liquidity via our API.
Use AI to create markets faster, smarter, and with greater accuracy.
AI-powered resolution that's fast, transparent, and unbiased.
Explore how we've helped brands scale engagement, boost performance.
VIEW ALL SUCCESS STORIESCommon questions about our prediction market AMM liquidity engine.