Prediction market signals: momentum, divergence, flow, and resolution risk

A prediction market signal is a rule that fires when live market data matches a specific pattern: price continuation, short-term reversal pressure, clustered large trades, or settlement ambiguity. A useful signal should name the pattern, expose the evidence, and say what to verify next.

The mistake is treating every signal as a forecast. Orrery does the opposite. It treats signals as a research queue. Momentum says a move is continuing. Divergence says the latest move is fighting the 24h trend. Flow says large trades have clustered. Resolution risk says the settlement path deserves inspection.

Momentum fires when the 1h and 24h probability changes point in the same direction and the short-term move is large enough to matter. It is strongest when the market is liquid, the spread is tight, and the move is not just a single thin print.

Momentum is not a guarantee that the move continues. It is evidence that the market is repricing in one direction across multiple horizons. The next step is to inspect the market page: source, trade flow, related markets, and whether the move is broad or isolated.

• Best used for: finding markets where attention is already concentrated.
• Weak when: liquidity is thin, spread is wide, or the market is close to settlement.
• Verify next: 24h volume, related markets, resolution source, and recent trades.

Divergence fires when the 1h move and 24h move have opposite signs. It is useful because it catches markets where the latest information may be conflicting with the older consensus.

A divergence can mean many things: a fresh data point, a liquidity gap, a temporary orderbook imbalance, or a market correcting a stale price. That ambiguity is why Orrery labels it as something to inspect, not something to trade.

• Best used for: finding markets where new information may have arrived.
• Weak when: the 1h move is tiny or comes from a single illiquid trade.
• Verify next: event siblings, source freshness, and whether whale flow confirms the move.

Flow looks at behavior rather than price alone. It fires when multiple large trades hit the same market in a recent window and lean toward the same outcome.

Large trades matter because they can reveal attention, urgency, or information-seeking behavior. But a whale trade is not automatically smart money. Some large traders are hedging, some are exiting, and some are simply wrong. The quality question belongs at the wallet layer.

• Best used for: detecting where capital is paying attention.
• Weak when: one wallet dominates the whole cluster.
• Verify next: unique wallet count, trade side, wallet history, and whether price moved after the trade.

Resolution risk is the signal that most protects users from false confidence. It fires when settlement may be messy: UMA is pending or disputed, the resolution source is ambiguous, the market is near expiry, or the orderbook is pinned before official confirmation.

This signal is not bullish or bearish. It says the market's rules deserve attention. In prediction markets, being right about the outcome is not enough if the market resolves according to a narrower rule than readers expect.

• Best used for: avoiding markets where the rule is harder than the headline.
• Weak when: the market has already settled with explicit upstream confirmation.
• Verify next: exact resolution source, UMA status, end date, and whether rail-pinned price has been mistaken for resolution.

A high-quality signal surface separates three labels. Evidence measures live strength. Backtest measures how the rule has behaved historically. Action names a verification step. Collapsing those into one score makes the interface feel confident, but it hides the thing users most need to know.

Orrery's action labels are intentionally verbs like monitor, inspect timeline, create alert, verify source, and ignore. There is no buy or sell label because the product is an intelligence layer, not a broker or advisor.