UOA for quant research

Unusual Options Activity for Quant Research

Define unusual-activity rules as testable SQL, measure hit rates over weeks of prints, then deploy thresholds to the real-time WebSocket without a vendor score.

For quant and systematic desks that treat UOA as a feature research problem. You own labels and thresholds. OptionData provides OPRA prints, chains, and 2.8B+ historical rows.

UOA product overview Historical SQL docs

From hypothesis to live UOA monitor

Codify rules, estimate false positives historically, attach chain context, then stream survivors.

State the rule in SQL

Example: premium ≥ $100k, DTE ≤ 14, repeated prints, optional |delta| bands. all queryable on RawOptionTrades.

Measure base rates

Count how often the rule fires per symbol/day over multi-week windows before wiring alerts.

Join chain context

Pull IV, OI, and bid/ask around event times so labels include liquidity and surface state.

Deploy to WebSocket filters

Translate surviving thresholds into server-side stream filters and online scorers.

What “unusual” can mean in research

Large premium vs recent symbol baseline
Short-dated clustering into a single expiry
Size relative to open interest (via chain join)
Repeated same-strike prints within a time window
AGGREGATED sweep-like consolidations
Custom desk labels. never forced by OptionData

Example: 30-day short-dated large-premium clusters

SELECT
  symbol,
  expiration_date,
  strike,
  put_call,
  sum(toFloat64(price) * size * 100) AS premium,
  max(iv) AS max_iv,
  avg(abs(delta)) AS avg_abs_delta,
  count() AS prints
FROM RawOptionTrades
WHERE date BETWEEN today() - 30 AND today()
  AND toFloat64(price) * size * 100 >= 100000
  AND dateDiff('day', date, expiration_date) <= 14
GROUP BY symbol, expiration_date, strike, put_call
HAVING prints >= 3
ORDER BY premium DESC
LIMIT 100

UOA research graph

See also the UOA product page, quant flow guide, and OI glossary.

UOA for quant FAQ

Does OptionData sell a proprietary UOA score?

No. It provides the data plane (stream, SQL, chains). Detection logic and labels are defined by your research.

Can I backtest UOA rules?

Yes. Historical SQL over stored option trades supports multi-day and multi-week rule evaluation with premium, Greeks, IV, and contract fields.

How do I avoid overfitting alert rules?

Hold out calendar weeks, track alert frequency per symbol, and require stability across AGGREGATED vs RAW if both modes matter to production.

Where do I start if I am new to UOA?

Start with the unusual options activity API overview, then return here for the research loop and SQL examples.