Prediction markets move faster than equities.
We capture the gap.
When a Kalshi or Polymarket contract reprices sharply, correlated equities take minutes to catch up. This engine detects velocity spikes — Δp/Δt exceeding threshold — and submits positions via Robinhood's agentic trading MCP before the gap closes.
Velocity, not probability.
Equity markets lag.
Prediction markets are purpose-built for rapid repricing. When new information arrives, contract probabilities update in seconds. Equity prices take minutes.
Δp/Δt > 0.15
A probability velocity exceeding 0.15 units per minute, confirmed by a volume spike, indicates genuine information arrival — not noise. Two conditions must hold simultaneously.
A 2-hour window.
Positions are held until the equity market reprices or 2 hours elapse — whichever comes first. The thesis is information diffusion speed, not prediction.
How it works.
Bring your own signal.
.prism is the connector format for the Velocity Signal Engine. Drop a .prism package into the connectors/ directory and the engine loads it automatically. Prediction markets, alternative data, news sentiment, order flow — if it produces a time series, it can drive a signal.
class MyConnector(PrismConnector):metadata = PrismMetadata(name="My Data Source",slug="my_source",source_type="prediction_market",transport="rest",auth_fields=["MY_API_KEY"],contract_slugs=["KXFED"],)async def start(self, tracker, mapper, handle_signal):while True:price = await self.fetch_latest_price()tracker.update(self.metadata.slug, PricePoint(timestamp=datetime.now(UTC),price=price, # must be 0.0 – 1.0volume=1,))await asyncio.sleep(self.metadata.poll_interval_seconds)
Live execution requires Robinhood agentic trading account access, currently in private beta. Set EXECUTION_MODE=live only after receiving access confirmation.