Proprietary quantitative market data
Proprietary datasets for overlooked market behavior.
Alphanume turns difficult-to-reconstruct signals, events, and point-in-time market states into research-ready datasets. Explore the rows first, understand the trading thesis, then decide what belongs in your process.
- Dataset catalog
- 27
- Construction
- Point-in-time
- Delivery
- Dashboard · REST · MCP
- Public preview
- No sign-in required
Dataset briefing
What it measures
Spot financing where a falling stock price can mean more shares issued, and size the raise against the company behind it. The feed covers ordinary PIPEs and structured private placements, with disclosed deal terms and a toxicity grade for each financing event.
The toxicity score grades financing structure, not future returns. A zero is not a clean bill of health, and an ungraded deal is not a safe one.
How traders use it
- Filter min_toxicity=3 to study market-linked financing, then inspect stated floors and discounts.
- Select event_kind=announcement for new agreements; review amendments separately to avoid counting every filing as a new raise.
- Compare gross proceeds with market cap, keeping equity-line commitments separate from funded amounts.
announce_date
8-K filing date in ET, YYYY-MM-DD. All date filters and default ordering use this date.
ticker
Issuer symbol; missing mappings can be filled later.
event_kind
announcement, closing, amendment_or_reset, termination, unstated. Separate filings can report different stages of the same deal.
structure
common, units_with_warrants, convertible_note, convertible_preferred, equity_line, sepa_prepaid_advance, other. Convertibles with warrants retain their convertible structure.
toxicity_score
Derived structural grade from 0 to 3, with 3 for market-linked pricing. Null means ungraded, not benign. See the scoring rules in the docs.
gross_proceeds_usd
Stated gross proceeds in USD, never inferred from principal or net proceeds. A facility commitment is not funded proceeds.
proceeds_pct_of_market_cap
Gross proceeds / market cap times 100. Null if either input is unavailable.
price_basis
fixed, variable_discount_to_market, lesser_of, or unstated. lesser_of takes the lower of a fixed and market-linked price.
warrant_coverage_pct
Investor common-warrant shares / shares_sold times 100, for common and units_with_warrants deals only. Excludes pre-funded and placement-agent warrants; null for convertibles, no common warrants, or missing counts.
record_id
Stable row ID and pagination tiebreaker.
accession_no
Public EDGAR accession number of the disclosure.
cik
SEC issuer identifier, retained across ticker changes.
company_name
Issuer name from filing metadata.
event_date
Stated event date. Month-only or year-only dates use the first day of that period; never inferred from the filing date.
closing_date
Stated closing date: can be expected on an announcement or actual on a closing disclosure.
securities_summary
Securities sold, as described in the filing.
committed_capacity_usd
Maximum purchase commitment for an equity line, prepaid advance agreement, or multi-tranche facility.
shares_sold
Stated securities count; common deals include shares underlying pre-funded warrants. Convertible counts can refer to notes or preferred shares, not common equivalents.
price_per_share_usd
Stated purchase price or initial/fixed conversion price, including the fixed leg of lesser-of pricing.
discount_to_prior_close_pct
100 times (prior close minus stated price) / prior close. Positive means a discount; negative means a premium. Uses an as-traded close before the day-specific event date, otherwise filing date. Null for missing or unsuitable per-common-share prices or unavailable price history.
variable_discount_pct
Discount to the variable market reference: 80% of VWAP is a 20% discount, stored as 20.
lookback_days
Trading days in the variable-price measurement window.
floor_price_usd
Stated minimum variable price in USD.
floor_exists_flag
1: stated floor; 0: explicitly no floor; null: unstated.
reset_provision
none_stated, price_reset_on_dilutive_issuance, weighted_average, or full_ratchet. Routine split adjustments do not establish a reset.
warrant_exercise_price_usd
Investor warrant exercise price; the lower price if series differ.
warrant_term_months
Investor warrant term in months; 60 means five years.
has_prefunded_warrants
1: pre-funded warrants sold; 0: stated securities exclude them; null: unknown.
original_issue_discount_pct
Stated original-issue discount in percent, not inferred from principal and purchase price.
investor_names_stated
Named purchasers, separated by semicolon and space. Null if identities are undisclosed.
investor_type_stated
institutional_investors, accredited_investors, existing_investors, single_institutional_investor, insiders_or_affiliates, related_party_or_officer, other, or unstated.
placement_agent
Named agents or financial advisers, separated by semicolon and space.
exemption_stated
section_4a2, regulation_d, regulation_s, multiple, or unstated.
registration_deadline_days
Stated deadline to FILE a resale registration, using calendar or business days as the filing counts them.
resale_registration_form
First linked resale registration form found within 180 days after the filing date.
resale_registration_date
Filing date of the linked resale registration, YYYY-MM-DD.
resale_effective_date
Effectiveness date of the linked registration, when available. Does not establish actual investor sales.
is_spac_related
1 for issuers classified under SIC 6770 (blank-check companies), otherwise 0.
concurrent_registered_offering_flag
1: concurrent registered offering; 0: no registered component; null: unknown.
terms_in_exhibit_flag
1: terms are in an unread exhibit and pricing is unstated; 0: terms in the body; null: unknown.
market_cap_at_announce_usd
Latest market cap strictly before announce_date, within 10 days. Coverage can lag for microcaps.
confidence
Quality signal from 0 to 1 for how faithfully the terms reflect the disclosure, not a probability of a price move.
first_seen_at
Filing timestamp used by the current storage code as an observation proxy, not verified API availability; pre-launch backfill is synthetic.
published_at
Projection timestamp in the served table. Reprojection can refresh this value; it is not an immutable historical availability record.
last_updated
Latest served-row update; used by updated_since. Timestamps use YYYY-MM-DD HH:MM:SS without a timezone suffix.
{
"count": 1,
"has_more": false,
"next_cursor": null,
"data": [
{
"record_id": 1001,
"accession_no": "0000000000-26-000001",
"ticker": "EXAMPLE",
"cik": 0,
"company_name": "Illustrative Company",
"event_kind": "announcement",
"announce_date": "2026-09-10",
"event_date": "2026-09-09",
"closing_date": null,
"structure": "convertible_note",
"securities_summary": "Privately placed convertible notes",
"gross_proceeds_usd": 2000000,
"committed_capacity_usd": null,
"shares_sold": null,
"price_per_share_usd": null,
"discount_to_prior_close_pct": null,
"price_basis": "variable_discount_to_market",
"variable_discount_pct": 20,
"lookback_days": 10,
"floor_price_usd": 0.25,
"floor_exists_flag": 1,
"reset_provision": "none_stated",
"warrant_coverage_pct": null,
"warrant_exercise_price_usd": null,
"warrant_term_months": null,
"has_prefunded_warrants": 0,
"original_issue_discount_pct": null,
"investor_names_stated": null,
"investor_type_stated": "institutional_investors",
"placement_agent": null,
"exemption_stated": "regulation_d",
"registration_deadline_days": null,
"resale_registration_form": null,
"resale_registration_date": null,
"resale_effective_date": null,
"is_spac_related": 0,
"concurrent_registered_offering_flag": 0,
"terms_in_exhibit_flag": 0,
"toxicity_score": 3,
"market_cap_at_announce_usd": 10000000,
"proceeds_pct_of_market_cap": 20,
"confidence": 0.9,
"first_seen_at": "2026-09-10 16:15:00",
"published_at": "2026-09-13 08:20:00",
"last_updated": "2026-09-13 08:20:00"
}
]
}Live rows are available with Pro.
Unlock the latest observations, complete history, REST access, and higher request limits.
Unlock with Pro →We test the data against what happened next.
Every study uses Alphanume observations as they were available at the time. Strong results, weak links, proxy limitations, and gross-versus-net caveats are reported on the same page.
- Datasets tested
- 12
- Confirmed cleanly
- 10
- Reported with caveats
- 2
- Charts published
- 36
Different datasets. Measurable effects.
- Stock Dilution
Median -14.9% at one month.
Across the reported study, 71% of dilutive filings were lower one month later (z = 8.9), with a stronger effect in smaller companies.
- Earnings Implied vs Realized
The implied move exceeded the realized move 64% of the time.
The published study reports a gross edge of 1.1% of spot per event (t = 10.4). Costs, slippage, and path risk still matter.
- IV/HV Premium
A 14.5-vol-point spread across reported quintiles.
The seller edge sorted monotonically from -9.4 to +5.1 vol points as implied volatility became richer versus realized volatility.
These studies establish historical market effects under the stated methodology. They are not expected returns, personalized advice, or a guarantee that an effect survives costs or future regimes.
Data you can defend in a research review.
- 01
Point-in-time by default
Dataset documentation distinguishes fixed observations from evolving deal records and filing histories. Check filing dates and availability timestamps so later information does not leak into an earlier test.
- 02
Constructed from primary sources
Alphanume turns raw filings, market observations, and model inputs into research-ready event histories and signals. Dataset pages state what came from the source and what Alphanume created.
- 03
Evidence attached
Each tested dataset links to the sample, period, test, result, and caveats. Nuanced findings remain visible instead of disappearing from the catalog.
- 04
One consistent interface
Use the same key across the dashboard, REST API, and hosted MCP server. Add a new signal without a new vendor integration.
Pro for your research. Enterprise for your team.
Pro
$349/moor $1,999/year, save 52%
Full history, live data, and internal live-trading rights for one researcher.
Enterprise
CustomSeats, SLA, custom delivery
Deploy across a team or product, with rights by contract.
Looking for the course? Alphanume Learn is separate.
$499 once for permanent course access, or included while subscribed to Pro. The five-lesson introduction is free with no account required. A Learn purchase does not include Pro or general-purpose API access.
Explore Alphanume Learn →The answers a buyer actually needs.
- Can I explore Alphanume before subscribing?
- Yes. Public dashboard previews, basic pricing calculators, documentation, and Proof studies remain available without a subscription. Preview availability and limits vary by dataset. Pro provides full data access, dashboard exports, and advanced calculators; Enterprise supports organizational requirements.
- What is happening to the Free data plan?
- New Free signups are closed. Our Free data plan retired September 6, 2026. Free API and MCP access has ended. Existing users can upgrade through their account. Learn's five-lesson introduction remains available without an account.
- Can I use Pro in a live trading workflow?
- Yes. Pro is the self-serve plan for one named user and may be used for that user's research, backtests, automation, and internal live trading. Choose Enterprise when you need multiple users or keys, redistribution or client-facing rights, custom delivery, an SLA, or organization-specific contract terms.
- What makes the datasets proprietary?
- Alphanume constructs datasets from primary sources and its own models rather than simply reselling a standard market feed. Each dataset page identifies the source inputs, the fields or labels Alphanume created, the observation-time policy, update cadence, and historical coverage.
- Does every dataset have proof?
- Every dataset has a documented methodology and point-in-time contract. Tested datasets also link to a public market-effect study. The Proof page reports strong and nuanced findings alike rather than labeling every result a success.
- How does agent access work?
- Add https://mcp.alphanume.com/mcp as a custom connector in Claude or Codex and sign in with your Alphanume account. There is no key to copy and no separate MCP subscription: sign-in resolves the key already on your account, so your tier and limits match the REST API exactly. Cursor, scripts, and CI use a key endpoint instead. The agent chooses the tools and parameters; Alphanume returns the same deterministic dataset rows available through REST.
- Are new datasets included in Pro?
- New standard Alphanume datasets are included in Pro as they launch. Separately licensed third-party data, custom datasets, bulk delivery, and organization-specific rights may require Enterprise.
- Are these trade recommendations?
- No. Alphanume provides data, models, and historical research for quantitative workflows. The published studies describe historical effects under stated assumptions; they are not personalized recommendations or guarantees of future performance.
Give your next strategy a differentiated input.
Explore the public dashboard preview, then access the full catalog with Pro. Every dataset ships under the same key, schema conventions, and point-in-time contract, with new releases included.