Sweep ~12,000 US equities and the top 100 cryptos for setups that match your shape. Compose any condition from 100+ queryable fields, then check the historical rhythm of every match — point-in-time signal fires going back up to all-time, per ticker.
GET /v2/scan + /v2/signals
Start with a SQL WHERE clause to filter the live universe. Pull historical fires for any signal × ticker to see whether the pattern shows up often or rarely. Compose with pre-computed indicators (rsi, macd, …) and named flags (breakout, gap_up_3pct, …) — same vocabulary across every endpoint.
-- small caps pattern-matching repeat breakouts
breakout
AND volume_unusual_2x
AND market_cap < 2_000_000_000| ticker | breakout | rel_vol | fires (5y) |
|---|---|---|---|
| RKLB | ✓ | 3.1× | 12 |
| IONQ | ✓ | 5.2× | 7 |
| ACHR | ✓ | 2.8× | 5 |
| BBAI | ✓ | 2.3× | 9 |
primitives you’ll use
Discovery sweeps the universe and confirms the pattern repeats. Three primitives combined.
ScreenersSQL filter against the live universe. Every field and flag composable, no fixed library.
See screeners →Signals50+ named flags plus historical fires per ticker. Confirm a setup is real, not a one-off.
See signals →IndicatorsRSI, MACD, ATR, Bollinger Bands as queryable columns. Compose into any setup.
See indicators →why this works
Not a fixed library of screens. Combine boolean flags (breakout, gap_up_3pct) with numeric columns (market_cap, rsi) using arbitrary SQL.
Every named signal has a point-in-time time series per ticker. See whether NVDA going RSI-oversold is a 2x/year event or a 20x/year event before you bet on it.
~12,000 US equities + top 100 cryptos. Find setups in small caps that an SPY-only stack would never surface.
No TA library to maintain. RSI, MACD, ATR, Bollinger Bands are columns in the schema, recomputed every minute.
comparison
Browser-based discovery, one ticker at a time
SQL across the universe + signal history
works with your agent
Hand the discovery loop to your agent. Same SQL grammar for the live scan and the historical confirmation, so the agent can iterate on dozens of setup hypotheses without swapping mental models.
ship it