Mercor names two finance listings, an investment firm listing at USD 50,000 to 800,000 and a hedge fund or quant firm listing at USD 25,000 to 1,150,000, both Mercor's own whole-deal claims, re-read 2026-11-17.
The SEC's insider trading texts, read in the eCFR on 2026-11-17, treat a duty of confidence as arising whenever a person agrees to keep information confidential, and the NYSE market data policy package dated 2022-03-21 contains no wording on AI training.
Every figure below is a dated claim from a named source, and investment and trading records are one sector in AI training data licensing.
How we checked this page: every clause and figure was compared with its primary source on the date shown, and the method is set out under how we source, date and check every figure.
What Investment Firms and Trading Firms Hold That AI Buyers Ask For
Two kinds of firm appear on buyers' pages: deal-side firms with closed-deal records, and trading firms with multi-year time series and code. Mercor describes each in its own listing, and the two listings ask for different records under different tests of ownership.
Closed-Deal Records: Data Rooms, CIMs, Models and Memos
Mercor's investment firm listing asks for records of closed deals: the data room, the CIM, the model and the investment memo for the same deal, with three or more years of history.
Mercor writes that "the most valuable material shows the full chain for one deal", and it requires the right to license deal and portfolio documents once anonymized.
It names private equity, private credit, boutique banks, quality-of-earnings providers, wealth managers and fund administrators as the firm types, and lists credit agreements with term sheets, quarterly portfolio reporting, banker deliverables, de-identified wealth client records, fund holdings, benchmarks, pricing and ESG data as nice to have.
Mercor lists documents under NDAs that forbid licensing as not a fit, and firms whose only records are their own operating finances are not a fit either. An old deal archive is a different decision from setting up a data room for an AI data licensing deal.
Proprietary Time Series and Trading Code
Mercor's hedge fund or quant firm listing asks for multi-year proprietary time series, or trading infrastructure code, that the firm owns and has the right to license for AI training.
The industries it names include proprietary trading and HFT, quant funds, broker-dealers and market-data vendors, and the tools it lists are kdb+, C++, Rust, FIX and Python.
Nice to have, in Mercor's wording, are order management systems, FIX, ITCH and OUCH handlers, low-latency engines, backtesters, research notebooks and backtest results. The listing also mentions sports odds histories, which this page does not develop.
Mercor's scoping questions for traders are "Which time series do you own, and how far back?" and "Is your trading code free of third-party licenses?" The questions are the buyer's own, and no license outcome follows from them.
Programs That Name Investment and Trading Records
Five program statements name investment, trading or finance records on their own pages, and none of them publishes a per-record price for them.
The table compares what each names or excludes, as of 2026-11-17, in alphabetical order, with the two Mercor listings and micro1's master terms re-read on 2026-11-17 and the other pages checked 2026-11-09. Each value is the program's own statement, and "not stated" does not mean allowed.
| Program | What it names, or excludes | Source |
|---|---|---|
| AfterQuery Data Partnerships | "finance records" and financial models; payroll, privileged legal material and credentials never used | afterquery.com/data-partnerships |
| Handshake AI Data Partnerships | "Finance & insurance" among 11 industries; "financial models, underwriting workflows"; "All data is scoped, de-identified, and confidential" | joinhandshake.com/ai/data-partnerships |
| Mercor, Hedge Fund or Quant Firm listing | multi-year proprietary time series or trading code the firm owns, right to license for AI training; not a fit: redistributed exchange feeds the firm does not own, data already public; deal range USD 25,000 to 1,150,000, "Depends on history and exclusivity"; 5 to 500 employees; published 2026-10-17 | api.data.mercor.com listing trading-firm |
| Mercor, Investment Firm listing | closed-deal data rooms, CIMs, models and memos; right to license once anonymized; 3+ years; not a fit: only own operating finances, documents under NDAs that forbid licensing; deal range USD 50,000 to 800,000, "Depends on deal count and completeness"; 5 to 200 employees; published 2026-10-17 | api.data.mercor.com listing investment-firm |
| micro1 Company Data Partnerships | "finance SOPs, reconciliations, reporting workflows"; payment card and financial account data excluded unless the Order Form allows (s.5.7, terms last updated 2026-11-12, re-read 2026-11-17) | micro1.ai/data-partnerships; data license master terms |
Mercor's two ranges are whole-deal claims that depend on deal count and completeness for the investment listing, and on history and exclusivity for the hedge fund listing, so they are never averaged or read as a price per record.
Ledgers, ERP exports and other finance records have their own page under financial records for AI training.
Records a Firm Owns Versus Market Data It Licenses
A trading firm's archive mixes what it generated with data it received under licence, and the licence on that data binds before any AI buyer's license does. The table sorts the record kinds by whose they are on the buyer's own wording, as of 2026-11-17.
| Record kind | Whose it is, on the buyer's own wording | What decides whether it can be licensed | Source |
|---|---|---|---|
| Multi-year proprietary time series the firm generated | "that the firm owns" (Mercor hedge fund listing) | the firm's right to license it for AI training | api.data.mercor.com |
| Redistributed exchange feeds | "the firm doesn't own" (Mercor, not a fit) | the exchange or vendor agreement; NYSE's policy package separates display use for the firm's own behalf and for its clients and names derived data for internal use, with no AI wording | api.data.mercor.com; nyse.com policy package |
| Data already public | not a fit (Mercor) | nothing to license | api.data.mercor.com |
| Trading infrastructure code | "that the firm owns" | whether it is "free of third-party licenses" (Mercor scoping question) | api.data.mercor.com |
| Closed-deal data rooms, CIMs, models, memos | the firm's records of deals | right to license once anonymized; NDAs that forbid licensing are not a fit; duty of confidence under 17 CFR 240.10b5-2(b)(1) | api.data.mercor.com; ecfr.gov |
| Client and wealth records | clients' information | written safeguards under 17 CFR 248.30(a)(1) | ecfr.gov |
Exchange and Vendor Licences on Market Data a Firm Does Not Own
Mercor's hedge fund listing lists redistributed exchange feeds the firm does not own and data that is already public as not a fit, so a firm's own time series and a feed it receives are different things.
NYSE's document "NYSE Proprietary Market Data", dated 2022-03-21 and published by Intercontinental Exchange, defines use categories: display use of real-time NYSE market information on the firm's own behalf, including the creation of derived data such as indices and financial products for internal use, and display use on behalf of its clients.
NYSE's pricing, policies and contracts page, read 2026-11-17, also lists a non-display declaration form. The policy text read contains no wording on AI training, so whether a given AI use counts as display, non-display, derived data or redistribution is a question for the exchange or vendor agreement, as the page on financial market data for AI: training licences vs access licences takes up.
Terms for AI use at other venues and vendors, among them CME, Nasdaq, LSEG and Bloomberg, are not recorded on this page, and nothing is inferred for them from NYSE.
Third-Party Code and Research Subscriptions Inside a Dataset
Trading code and research notebooks can carry other parties' work: third-party libraries, vendor APIs and subscribed research. Mercor's question "Is your trading code free of third-party licenses?" points at the same issue.
Open-source terms can attach to a codebase, as described under open-source license contamination when licensing a codebase for AI training. The terms of subscribed research and vendor reports on AI use are not recorded here, and the page names no research vendor and states no clause.
Material Nonpublic Information and Confidentiality Come Before Any License
Deal records and trading records can hold material nonpublic information, and the SEC's own texts show that duties of confidence attach to such information whenever a person agrees to keep it confidential.
The table lists the official texts read, as of 2026-11-17, from the eCFR as of 2026-11-10 and the 2023 edition of the U.S. Code.
| Instrument | Wording relevant here | Source |
|---|---|---|
| 17 CFR 240.10b-5 | unlawful to employ a device to defraud, make an untrue statement of material fact or omit one, or engage in a course of business operating as fraud, "in connection with the purchase or sale of any security" | ecfr.gov |
| 17 CFR 240.10b5-1(a), (b) | trading on the basis of material nonpublic information "in breach of a duty of trust or confidence" is covered; a trade is on the basis of it if the person "was aware of" it; insider trading law "is otherwise defined by judicial opinions" | ecfr.gov |
| 17 CFR 240.10b5-2(b)(1) | a duty of trust or confidence exists "whenever a person agrees to maintain information in confidence" | ecfr.gov |
| 15 U.S.C. 80b-4a (Advisers Act s.204A) | every adviser must "establish, maintain, and enforce written policies and procedures reasonably designed" to prevent misuse of material, nonpublic information | govinfo.gov |
| 17 CFR 243.100(b)(2)(ii) | Regulation FD's disclosure duty does not apply to a disclosure "to a person who expressly agrees to maintain the disclosed information in confidence" | ecfr.gov |
| 17 CFR 248.30(a)(1) | every covered institution must "develop, implement, and maintain written policies and procedures" safeguarding customer information | ecfr.gov |
What the SEC Texts Say About Confidence and Awareness
Rule 10b-5 prohibits fraud, deceit and untrue statements of material fact in connection with the purchase or sale of any security, and the SEC's later rules define when trading on nonpublic information falls inside it.
Rule 10b5-1(a) covers a purchase or sale on the basis of material nonpublic information in breach of a duty of trust or confidence, and 10b5-1(b) treats a trade as on the basis of it if the person was aware of the information.
The rule adds that the law of insider trading is otherwise defined by judicial opinions construing Rule 10b-5.
Rule 10b5-2(b)(1) says a duty of trust or confidence exists whenever a person agrees to maintain information in confidence.
Section 204A of the Advisers Act, 15 U.S.C. 80b-4a, requires advisers to keep written policies reasonably designed to prevent misuse of material nonpublic information, and Regulation FD, 17 CFR 243.100(b)(2)(ii), exempts disclosure to a person who expressly agrees to keep it confidential.
The texts read do not define "material" or say whether an anonymized dataset is itself material nonpublic information, so the page states no answer to that question.
Deal Rooms, NDAs and Client Records
A deal NDA is a promise to keep information confidential, and Mercor's investment listing treats documents under NDAs that forbid licensing as not a fit, while requiring the right to license once anonymized.
The NDA terms are the seller's first check, and the clause itself is covered under NDAs before sharing data with an AI buyer.
A data room usually holds counterparties' information, so consent and confidentiality clauses apply to it.
Handshake AI states "All data is scoped, de-identified, and confidential", and Mercor states "No AI lab buyer or third-party vendor ever sees your raw data", both checked 2026-11-09. These are buyers' claims, and they do not release a duty the firm owes.
Client records add a second layer: Regulation S-P requires every covered institution to keep written safeguards for customer information under 17 CFR 248.30(a)(1), and the privacy rules for such records sit under GLBA and licensing financial customer data for AI.
Benefits, Risks, Downsides and Alternatives for an Investment Firm
Licensing can earn income from records a firm already keeps, but the exchange, vendor, confidentiality and client-privacy limits shrink what can go, and not licensing is a valid outcome.
Benefits
- Mercor states whole-deal ranges for both listings, as claims dated 2026-11-17, that depend on deal count, history and exclusivity.
- Mercor states that it covers extraction, anonymization and transfer costs, checked 2026-11-09.
- The firm licenses a copy and keeps its own records.
Risks
- Licensing data received under an exchange or vendor agreement, where the agreement decides.
- Breach of a deal NDA, and a duty of confidence under 17 CFR 240.10b5-2(b)(1).
- Missing the written safeguards for client information under 17 CFR 248.30(a)(1).
Downsides
- Scoping work to separate own, licensed and public data.
- micro1's master terms grant a worldwide, nonexclusive, irrevocable, perpetual license in s.1.2, re-read 2026-11-17.
- A trained model cannot be told to forget what it learned.
Alternatives
- License only records the firm owns, such as its own research notes or procedures.
- License finance SOPs without deal files, the kind of material micro1 names.
- Keep the archive for in-house use.
- Not licensing at all, among the alternatives to selling data, including not selling at all.
Where Investment Licensing Meets the Seller Path, the Law and Other Industries
Investment firms share client secrecy with law and accounting firms and share the financial-privacy rules with banks, and the industry index sets out how each sector's limit differs.
This page is one entry in data licensing by industry, next to banks and financial firms licensing data, accounting and tax firms licensing data, consulting and advisory firms licensing client deliverables and law firms licensing data.
The full map is AI training data law by jurisdiction, with trade secrets under the Defend Trade Secrets Act and the EU trade secrets directive, and market data licences under financial market data for AI: training licences vs access licences. This page is information, not legal advice.
- Run a data sale readiness check on the records.
- Review counterparty terms under customer consent and confidentiality clauses.
- Remove client data through de-identification before licensing data.
- Work through choosing a buyer for your data.
- Negotiate the clauses of AI data licensing agreements.
Buyer and Law Pages Named on This Page
Each program named above has its own page with dated terms and sources.
- AfterQuery data buyer profile
- Handshake AI data buyer profile
- Mercor data buyer profile
- micro1 data buyer profile
Frequently Asked Questions
Can a hedge fund license exchange data to AI companies?
Only as the exchange or vendor agreement allows: Mercor lists redistributed exchange feeds the firm does not own as not a fit (re-read 2026-11-17), and NYSE's policy package (2022-03-21) has no AI wording.
How much do AI companies pay investment firms for data?
Mercor states USD 50,000 to 800,000 for investment firms and USD 25,000 to 1,150,000 for hedge funds and quant firms, whole-deal claims re-read 2026-11-17, not prices per record.
Valuation methods are set out under what company data is worth to AI companies.
Is a dataset made from closed deals material nonpublic information?
The SEC texts read define the duty and awareness but not materiality for a dataset, so this page states no answer; the firm's compliance policies decide.