A price-action API
“The last 10 times the chart looked like this, 7 went up.”
Pattern-matching on shape. Says nothing about why the move happened or what triggered it.
26M+ news records20+ years of dataZero look-ahead
Every market event, labeled with what actually happened.
20 years of realized price reactions and 93 features per event. Measured, not guessed — built for quants, AI teams, and agents.
Add to Claude — data insights in-chat
No card · 100 free queriesclaude mcp add news-quantified
26M+
point-in-time news records
2006–2026
20-year continuous dataset
93
engineered features per event
Who it’s for
One dataset. Four ways in.
For traders & investors
Check the historical base rate of an event class before acting on it — day-0 reactions, 5-day drift, reversal rates, per ticker.
For institutions
License the reference event-reaction dataset: 26M+ news records, peer-reviewed methodology, a decade of continuous institutional use.
For public companies
Know how the market treats announcements like yours before you announce — every press release scored against 20-year baselines.
For partners
Publisher partnerships and dataset licensing: your news catalog, outcome-labeled with measured market reactions.
AAPLQ3 EPS beat +7.4%+2.34%MSFTGuidance raised+1.88%NVDAAnalyst day, TAM+6.71%META8-K disclosure−0.42%TSLAProduction miss−4.50%AMZNAWS reaccel+3.12%GOOGLAntitrust ruling−1.27%JPMNII guide up+2.05%NFLXSub adds beat+5.43%AMDMI300 ramp+4.18%INTCFoundry delay−3.61%BAFAA cap held−2.14%
The question only we answer
Price-action tells you the chart rhymed. We tell you the event repeated.
News Quantified
“The last 400 times this event hit — earnings beat, downgrade, 8-K, insider buy, rumor, guidance — here is the actual reaction distribution.”
Ground-truth labels computed in-house, anchored at time-of-news, measured across pre / main / post sessions. No price-data vendor has 20 years of event-to-reaction labels.
What you can build
One dataset, four jobs.
Backtesting & research
Pull 20 years of point-in-time labeled reactions to backtest event-driven strategies with zero look-ahead bias.
Agent grounding (MCP)
Give an AI agent ground-truth event reactions over MCP — answer “what happens after this event” with real distributions, not guesses.
Signals & features
93 engineered features per event — see the realized reactions they label, per ticker and per news event.
Event monitoring
Track labeled events and their reactions as they happen — a live, queryable watchlist of market ground truth.
Explore the data
20 years. Millions of labeled events.
Classified events carry their realized reaction across three sessions, point-in-time with zero look-ahead. Browse the dataset by event type and by sector.
Apple (AAPL)
How AAPL actually reacts to news and analyst actions — realized, point-in-time moves per event.
NVIDIA (NVDA)
The measured NVDA reaction history — 5-minute and to-close moves, per headline.
Tesla (TSLA)
Every labeled TSLA event with the move the market actually made.
By event type
The six measured analyst event classes — and why the class label does not tell you the direction.
By sector
Every measured sector, ranked by how far a typical name's reactions actually span.
All tickers
Browse every name with a measured reaction history and per-event news pages.
Live today: earnings and analyst actions. SEC filings are expanding into the same labeled format.
One call, the whole picture
The whole picture, in one call.
An agent asks one question: given this event, on this name, what is the distribution of outcomes? The answer is the ground truth across every comparable event in the dataset — returned with its spread and its leakage guarantee, not a single point estimate that hides the variance.
Live query · news_reaction
Point-in-time ✓ · zero look-ahead
The agent's tool call
{
"tool": "news_reaction",
"symbol": "AAPL",
"event_type": "upgrades",
"lookback": "3y",
"as_of": "event_ts" // no look-ahead
}Point-in-time enforced · zero look-aheadEvery feature is computed from data that existed at the event timestamp. No restatement, no survivorship, no future leakage. The same guarantee your compliance team asks vendors for, enforced at the query.
news_reaction · ground truthn = 68 events
Real sample · pulled 2026-06-25
+0.36%
median 30-min reaction
middle 50% : -0.07% to +1.22%
middle 50% : -0.07% to +1.22%
upgrades · AAPL · last 3 years · horizon 30-min
5-min+0.10%
30-min+0.36%
60-min+0.30%
to close+0.66%
% positive · 30-min66%
events measured68
median to close+0.66%
One name, one event class. Query the full cross-section — every name, every event type — in the dataset.
Why it can't be rebuilt
You cannot backfill 20 years of point-in-time reactions.
01
Point-in-time, not reconstructed
You can't recompute how a stock reacted to historical news after the fact. The labels were captured as events happened — accreted continuously since 2006.
02
The label is the IP
The proprietary value is the in-house reaction computation, not the news text. It is source-independent: we monetize derived reaction labels, not raw newswire bodies.
03
A new entrant starts today
Anyone copying this begins labeling now and waits a decade for depth. The moat is time, and it compounds every trading day.
Independently validated
A peer-reviewed study built on ten years of News Quantified data — independent, and it validates the exact thing we sell.
Levi, Livnat, Zhang & Zhang (2016) — researchers at Tel Aviv, NYU Stern, Rutgers, and UC Berkeley — found that on news days, extended-hours returns positively and significantly predict the next regular session and the quarter-long drift. The realized reaction even subsumes the earnings surprise itself.
20-year dataset · millions of labeled events · 93 engineered features, every one point-in-time, zero look-ahead, no restatement, no survivorship.
Read the study →SSRN 2809950 · Independent academic work, study window 2006–2015. The data contains validated predictive signal — not a promise of tradable P&L.
Pricing
One credit. MCP or REST. Scale as you grow.
NQ delivers news and corporate events with ground-truth reaction context — not a market-data feed. One credit balance and one rate limit, used however you call us. The latency lever is how fresh your event delivery is.
Free
$0forever
Build against 20 years of labeled events.
No card. 100 credits refresh monthly.
- 100 credits/month — no credit card
- Every lookup tool, on prior-day & historical events
- 1 year of ground-truth reaction context
- Full integration sandbox — build & test end to end
- No current-day events (delivery starts on Builder)
Builder
$49/mo
Today's events, in your product.
- 10,000 credits/mo · overage $0.010/cr
- Current-day events, 15-minute delayed
- The full Pro analytics tool set
- 5 years of reaction context
- Commercial use
Pro
$249/mo
Low-latency events and fund-grade analytics.
- 75,000 credits/mo · overage $0.008/cr
- Current-day events, low-latency delivery
- Full self-serve history depth
- Metered fund-flagship tools (drift, decay, block flow)
- 300 req/min · priority support
Scale
$1,500/mo + usage
Embed NQ in your platform.
- 500,000 credits/mo · overage $0.006/cr
- B2B2C resale rights with attribution
- 1,000+ req/min, higher stream concurrency
- SLA-lite, dedicated support channel
Enterprise / Fund
Contact sales
The full labeled history, the latency data, training rights.
- FUND-ONLY tools: provider attribution, block screen, signal scan
- Lowest-latency delivery + raw first-touch & multi-provider latency fields
- Full 20-year labeled history, bulk export
- AI-training and redistribution rights
- Custom SLA, named support
Compare all features
Swipe to compare plans →| Free$0 | Builder$49/mo | Pro$249/moMost popular | Scale$1,500/mo+ | Enterprise / FundCustom | |
|---|---|---|---|---|---|
| Usage & limits | |||||
| Monthly credits | 100 | 10,000 | 75,000 | 500,000 | Custom |
| PAYG overage | — | $0.010 | $0.008 | $0.006 | Negotiated |
| Rate limit (req/min) | 5 | 60 | 300 | 1,000+ | Dedicated |
| Events & history | |||||
| Current-day events | — | 15-min delay | Low-latency | Low-latency + priority | Lowest latency |
| Historical events | ✓ | ✓ | ✓ | ✓ | ✓ |
| Reaction-context lookback | 1 yr | 5 yr | Full | Full | 20 yr |
| First-touch latency fields | — | — | — | — | ✓ |
| Multi-provider latency fields | — | — | — | — | ✓ |
| Tools | |||||
| Lookup tools | ✓ | ✓ | ✓ | ✓ | ✓ |
| Pro analytics tools | — | ✓ | ✓ | ✓ | ✓ |
| Fund-flagship tools | — | — | Metered | Metered | Full |
| FUND-ONLY tools | — | — | — | — | ✓ |
| Rights & access | |||||
| Bulk export | — | — | — | — | ✓ |
| Commercial use | — | ✓ | ✓ | ✓ | ✓ |
| B2B2C resale | — | — | — | ✓ | ✓ |
| AI-training rights | — | — | — | — | ✓ |
| MCP + REST | ✓ | ✓ | ✓ | ✓ | ✓ |
| Support | |||||
| Support | Community | Priority | Dedicated channel | Named |
What a credit costs
Cost per call is fixed by the tool's class — identical on MCP and REST. methodology_search is always free.
Lookup / customer-face1–2 crstock_snapshot · news_reaction · top_movers · analyst_consensus
Pro analytics3–5 crsmart_money · compare_reactions · earnings_drift · novelty_split
Fund-flagship8–10 crevent_block_flow · post_event_drift · reaction_decay
Questions
Do you provide real-time stock prices?
No. NQ delivers news and corporate events with ground-truth reaction context and inference. We are not a market-data feed. What scales across tiers is how fresh the event delivery is and how deep the reaction history goes.
What counts as a credit?
One tool call. Lookup tools cost 1–2 credits, analytics tools 3–5, fund-flagship tools 8–10. The exact cost is shown per tool and is identical on the MCP and the REST API.
Why can't I see today's events on Free?
Free is for building and evaluating against historical events. Current-day event delivery starts on Builder ($49/mo).
Do credits roll over?
Monthly credits reset each cycle. Annual plans bill upfront, include two months free, and roll unused credits forward within the term.
What happens when I run out?
Free stops at the cap. Paid tiers continue at the per-tier PAYG overage rate. You can set a spend cap.
Can I export the full history?
Bulk export and the full 20-year labeled history are Enterprise. Self-serve tiers are metered, per-call access.
MCP or API: does my plan cover both?
Both. One credit balance, one rate limit, used however you call us.
Can I resell NQ data inside my product?
Scale and Enterprise include B2B2C terms. Builder and Pro are for your own use.
From the research desk
Read the work behind the data.
The academic study
An independent, peer-reviewed paper (Levi, Livnat, Zhang & Zhang) built on ten years of NQ data — the reaction predicts the drift, and beats the surprise.
Reaction histories by ticker
How each name actually reacts to news — realized, point-in-time moves across labeled events, with a page per news event.