A plain explainer, for people who want to understand before they issue.

A Hedge policy is a small basket of Polymarket prediction-market shares that we buy on your behalf when you tell us what you're worried about.

A Polymarket share is a contract. It says: "On a specific future date, this will be worth exactly $1 if event X happens, or $0 if it doesn't."

When you tell Hedge "I'm worried about Bitcoin crashing," we look at every live Polymarket market and find the ones whose payout correlates with your concern. Then we size positions across them so your premium buys you the most coverage we honestly can.

The "policy" is just a typeset wrapper around those positions: a policy number, a coverage statement, a payout scenario table. But underneath it's literal, contractual claims on Polymarket markets settled by an on-chain oracle.

Two layers of work happen behind every policy you see:

1. Live market data

We sync Polymarket's full open-market universe every minute. Prices, liquidity, end dates, descriptions, everything. When you submit a concern, we have a fresh snapshot of every market that's actually tradable right now.

2. An AI relevance check

We use AI (specifically OpenAI's text-embedding-3-small and gpt-4o-mini models) in two places:

Embedding search. Every market gets a 1,536-number vector that captures what it's about semantically. When you type "Bitcoin crashing," we find the markets whose vectors are nearest to that phrase. This is how we know that "MicroStrategy sells BTC" is more relevant to your concern than "NBA Finals".

Relevance verifier. Embeddings are good but imperfect. When we first tested it, the system matched "Bitcoin crashing" to "EU dissolves" because both are about catastrophe. So we added a second AI pass that asks gpt-4o-mini, one market at a time:"Does the YES side of this market actually pay out if the user's concern fires? Yes or no." Strict by design. False positives are worse than false negatives: a wrong basket would be far more harmful than a small basket.

3. Pure math, no AI

The sizing of each position (how many shares to buy, what the coverage adds up to, what the payout scenarios are) is deterministic arithmetic. AI helps us find the relevant markets; math decides how much to buy. We do this on purpose: AI is good at "is this related," bad at "is this fair pricing."

Honest answer: most $50 hedges aren't economically meaningful. A $50 premium typically buys you $50-150 of coverage. That's not going to save you if a real disaster hits.

So why issue one? Three real reasons:

You want to prove the mechanic on your wallet. A $50 demo policy lets you see the loop end-to-end: premium paid → positions opened → policy in dashboard → market resolves → payout (or expiry). This is what we recommend for first-time users. Treat it as paid tuition for understanding a primitive.

You're hedging a tail risk where leverage is high. If a market trades at $0.03 (3% probability), every $1 of premium buys ~$33 of coverage. So if you're worried about something the crowd thinks is unlikely (Fed at 7%, a recession declared, a specific country in conflict), small premiums can buy meaningful coverage. That's the leverage of insurance: cheap premiums for high-impact, low-probability events.

You're building a multi-leg basket. A $50 single-position policy is small. A $200 basket spread across three correlated markets is real protection. As Polymarket's universe grows and your premium grows, the leverage compounds.

Hedge is built on Polymarket. Polymarket only has markets that people actively trade. That means there are real concerns Hedge can't price right now:

Niche or hyper-specific concerns. "Fed raising rates above 7% before my mortgage closes": Polymarket has Fed-cut markets, but not Fed-above-7% markets (the current cycle is cutting). When this happens, we'll honestly tell you we can't construct the hedge and surface the closest available presets instead.

Things insurance already covers cheaper. Health, life, home: buy real insurance. Hedge is for the gaps insurance leaves.

Concerns about your own actions. Markets resolve on public events ("Fed cuts rates," "Trump wins"), not on private outcomes ("I get laid off"). The best we can do is hedge a correlated public event, e.g. an NBER recession declaration as a proxy for your job risk.

If you need guaranteed coverage at a fixed price. A Hedge policy is a basket of market positions; the market prices move. What you actually buy at issue time is locked, but if you wait to buy, the basket may have shifted. We never promise a fixed quote beyond the moment you see it.

Polymarket isn't a casino to us. It's a place where you back your sentiment with real weight, and build a public record of whether your views hold up. That's a productivity primitive: a way to make your thinking accountable, to convert opinions into evidence.

Augury Hedge is the practical end of that primitive. When you have a real concern about something in your life, instead of just worrying, you can put a small amount of money behind a defensible view and get paid if you turn out to be right. That's the mechanic. It's the same logic as insurance: pay a small premium up front, get a defined payout if the trigger fires.

Augury Hedge is not insurance. Augury is not a licensed insurer. We're a builder app that constructs portfolios of Polymarket prediction-market positions on your behalf. Polymarket positions are settled by UMA's optimistic oracle and pay out only as the underlying markets resolve. You may lose your full premium. Markets may be illiquid at resolution. Augury earns a small fee on each position routed through its builder code. Polymarket's own availability varies by jurisdiction, and that's between you and them.