the prediction
market for the
AI era

the most efficient market microstructure, with fully trustless, verifiable execution.

testnet opening soon

will Anthropic require KYC for Claude Code by end of 2026?batch 1 / 5
batch window · 5s5.0s
0.750.500.250.500.710.660.380.49news ↑news ↓indicative12345
indicative · 0.50

traders compete on price, not speed

orders don't execute first-come-first-served. they collect over a short window and clear at one fair price — so trading rewards the best information and pricing, not the fastest reaction speed.

batch window · 5s5.0s
0.800.700.600.500.40news ↑clears · 0.68
indicative · 0.50

FAQ

Existing venues did their job — they popularized the idea and grew the audience. Now they’re drifting into casinos with a thin slice of genuinely important markets, sitting under strict American oversight. Sybil takes the other path. Building a prediction market from scratch, properly: the most efficient market microstructure, verifiable trustless settlement, privacy by default, and maximum capital efficiency. To dive deeper, check our substack ↗ and research page ↗.

Almost every existing prediction market is an AMM or a CLOB — both a poor fit for prediction markets, and a nightmare for market makers, who survive only on heavy liquidity rewards. In frequent batch auctions, orders collect over a short window and clear at one uniform price, radically lowering the bar for market making and removing the speed race and news sniping. To know more, check our articles ↗ and math paper ↗.

Sybil is a general prediction market, but it starts where competitors are weakest — publicly important markets that are underserved, or simply not viable on a CLOB: AI markets, and long-tail political events outside the US.

Testnet is planned for mid-summer. Mainnet follows in autumn.

still early

testnet · mid-summer 2026mainnet · autumn 2026