Mine years of point-in-time history for setup patterns and rare-event signatures. Same SQL screens live or historical — no lookahead, no separate research stack. Custom signal definitions and redistribution rights on Enterprise.
GET /v2/scan?asof=
Loop the same query across a date range. Every match is point-in-time, with no leakage or survivorship bias. Pull the forward returns from the time series endpoint, build a triggers table, validate the edge before sizing the trade.
-- alpha research: same SQL, point-in-time
GET /v2/scan
?q=pct_from_52w_high > 25
AND rsi_oversold
AND volume_unusual_2x
AND momentum_strong_up
&universe=research_universe_q4
&asof=2025-09-30| ticker | pct_from_52w_high | rsi_14 | relative_volume |
|---|---|---|---|
| PLTR | 28% | 29.1 | 3.4× |
| SOFI | 34% | 28.4 | 2.8× |
| U | 41% | 27.6 | 2.1× |
| RBLX | 32% | 29.5 | 2.6× |
| NET | 26% | 29.7 | 2.4× |
| CRWD | 29% | 27.1 | 2.9× |
what you can build
Point-in-time backtests across the historical window. Build triggers-and-outcomes tables in one HTTP loop.
See how →Name your research universes; reference them by slug. Run any SQL screen across any subset.
See how →Enterprise: bring your own signal definitions; we compute and serve them alongside ours.
See how →comparison
$50k+ per seat, internal engineering
Point-in-time SQL, no engine to build
works with your agent
Hypothesis-test alpha programmatically without staffing a quants team. Point-in-time history queryable with ?asof= means agents can replay any setup against any date and trust the result.
ship it