Why prediction markets need a verification layer

Most people read prediction markets as a number. A market is 72 percent, 18 percent, 99 cents, or 4 cents. The number is easy to screenshot, easy to repeat, and easy to mistake for a complete answer.

But the price is only one layer. A market can move sharply on thin liquidity, sit near expiry, be pinned near a rail without official settlement, or depend on a source that is narrower than the headline suggests. Without that context, the same price can mean very different things.

The first Orrery principle is simple: price is not settlement. A market trading near 1 cent or 99 cents can look final before the upstream source, oracle, or exchange status has actually resolved it.

That distinction matters because prediction markets are not just forecasts. They are contracts with deadlines, resolution wording, source dependencies, dispute paths, and status transitions. Reading only the price hides the exact details that decide whether the market should be trusted yet.

• Near 100 cents can mean strong market confidence, not official finality.
• Expired can mean the event window ended, not that settlement has completed.
• Disputed or source-pending states can change what evidence matters next.
• A screenshot without status context can spread a false sense of certainty.

The biggest mistakes often happen when a market changes what kind of information it is sensitive to. A live market may care about news, volume, and related-market movement. A near-expiry market may care more about source wording and cutoff time. An expired but unresolved market may care almost entirely about settlement mechanics.

Participants can keep pricing the market as if it were still in the old state. That is where verification becomes operational. The question is not only where the market is now. The question is what changed about what the market should pay attention to.

• Live to resolution-sensitive: verify deadline, source, and whether the event window is closing.
• Near-expiry to expired: separate outcome belief from official settlement state.
• Rail-pinned to resolved: confirm upstream status before treating price as final.
• Disputed or source-pending: map evidence directly to the resolution criteria.

Orrery turns prediction-market movement into a ranked verification workflow. It asks what moved, why it matters, what can go wrong, and what to verify next. That is different from showing odds alone, and it is different from selling a directional call.

The product surfaces the same trust frame across Home, Markets, Daily Brief, Signals, Research Queue, Alerts, share cards, and APIs. A user should be able to move from a market card to a source check to an alert without losing the distinction between observation, evidence, and settlement risk.

• Market movement: probability change, volume, liquidity, spread, and related-market context.
• Source context: resolution wording, source extraction, expiry, and settlement path.
• Risk context: UMA/source pending, stale expiry, rail-pinned states, and confidence labels.
• Next action: inspect timeline, verify source, create alert, watch only, or ignore as too noisy.

Prediction-market tools often make large trades feel like a shortcut. A big wallet bought, therefore the market must know something. That framing is dangerous. Large trades can reflect hedging, exiting, liquidity taking, experimentation, or simply being wrong.

Orrery treats wallet and trade data as context. It can help answer where attention is moving, but it should not be dressed up as a copy-trading instruction. Activity is not profitability, and size is not evidence by itself.

• Show sample confidence before implying a wallet pattern matters.
• Separate activity, concentration, and historical behavior from outcome accuracy.
• Label unknown context clearly instead of turning trade size into a claim.
• Keep wallet pages research-first and never frame them as copy-this-wallet feeds.

A verification layer becomes useful when it is part of a daily workflow. Orrery's Daily Brief is built as a morning control room: what moved, what became risky, what deserves source checks, and what can probably be ignored for now.

That matters because prediction markets are noisy. The user does not need every market. The user needs a small set of markets where the next verification step is clear. A good brief should reduce attention cost without pretending to remove judgment.

AI agents can call market APIs quickly, but raw data alone is not enough. If an agent sees only price and volume, it can repeat the same mistakes as a human reader: treating rail-pinned prices as resolved, confusing similar contracts, or ignoring source wording.

That is why Orrery's agent layer is built around structured decision cards, sources, freshness, risk labels, and not_trade_advice metadata. The agent should know what to inspect next, not receive a disguised trade instruction.

• Free discovery endpoints help agents find markets and brief previews.
• Paid depth endpoints explain movement, resolution risk, and suggested next calls.
• Every useful agent response should preserve sources, freshness, and safety language.
• A good agent can verify context faster, but it should not convert context into buy or sell instructions.

Orrery is not a broker, exchange, market maker, copy-trading product, or betting tip sheet. It does not execute trades, custody funds, or recommend buying and selling. That boundary is not cosmetic; it is central to the product category.

The goal is to make prediction-market information harder to misread. When a price moves, Orrery should help a human, team, or agent ask the right next question: can this price be trusted yet?