Yes, for data and code the firm owns. AI buyers ask trading firms and data providers for multi-year proprietary time series, such as order-book, liquidity and risk data or sports odds and outcomes, or for trading infrastructure code. The firm also needs the right to license it for AI training. Redistributed exchange feeds and public data are not a fit, and exchange data agreements are the check that stops most sets.
- Two assets qualify: proprietary time series, or infrastructure code the firm owns.
- A feed you recorded from an exchange is the exchange's data under your data agreement.
- Order management, protocol handlers, low-latency engines and backtesters add value.
- History and exclusivity are what the published range depends on.
What trading data and code do AI buyers ask for?
Either long time series the firm owns, or the code it trades with. A public listing for hedge funds and quant firms (one buyer's public listing, read 15 November 2026) sets two requirements: "Multi-year proprietary time series, or trading infrastructure code, that the firm owns" and "The right to license it for AI training".
It describes the data as "long proprietary time series, such as order-book, liquidity and risk data or sports odds and outcomes", and lists what adds value: "Order management systems, FIX/ITCH/OUCH handlers, low-latency engines and backtesters", "Research notebooks and backtest results" and "Sports odds and outcome histories".
| Asset | Form buyers ask for | Ownership question |
|---|---|---|
| Order-book and liquidity series | Timestamped, documented schema, several years, gaps listed | Was it built from a venue's feed under a data agreement? |
| Risk and position data | The firm's own fills, positions and risk snapshots over time | Client and counterparty names removed? |
| Sports odds and outcomes | The lines the book set, with timestamps and settled results | Which inputs came from a league or official data supplier? |
| Trading infrastructure code | Repositories with history: order management, protocol handlers, engines, backtesters | Any third-party or open-source licence inside? |
| Research notebooks and backtests | Notebooks with the results they produced | Which licensed data sets do they embed? |
Nothing on this page is investment advice. It is about records a firm already holds, not about trading.
Does exchange data you recorded count as yours?
Usually not. The listing excludes "Redistributed exchange feeds the firm doesn't own" and "Data that is already public". Recording a feed for years gives you a valuable file, but the data in it is still licensed to you by the venue under your data agreement.
Exchange policies say this in terms. One exchange group's market data policies state that "Before a firm can redistribute Historical Data to non-Affiliates, the firm must first sign a Data Agreement", and list "Training or operating a Large Language Model trained on the Data" as an example of non-display use, a category they treat separately from showing data on a screen (Cboe Market Data Policies). Other venues have their own terms. Read the agreement for every feed in the set.
What is more likely to be yours: your own orders and fills, the quotes you made, your internal marks and risk numbers, and data from a venue you operate. Series derived from a venue's feed sit in between, because derived-data terms differ by venue. List the inputs to each series before you call it proprietary.
Is trading code worth anything once a strategy is retired?
It can be. The listing treats infrastructure code as a route of its own: firms that own their trading infrastructure code qualify without any data. The code it names is plumbing, not signals: order management, protocol handlers, low-latency engines and backtesters.
Handlers are a good example of what is yours. FIX "is a global, open standard that enables financial institutions to communicate trading information quickly, accurately and securely" (FIX Trading Community). The standard is public. Your implementation of it, tuned over years and tested against real venues, is your own work.
The listing's scoping questions include "Is your trading code free of third-party licenses?", so a dependency and licence scan comes first. Trade secret law also expects care: information is protected only if "the owner thereof has taken reasonable measures to keep such information secret" (18 U.S.C. 1839). Remove credentials, venue session details and live parameters, and license under confidentiality. What buyers check in a repository is covered in the source code guide; old systems are covered in the legacy code guide.
Which firms qualify?
The listing covers proprietary trading and high-frequency firms, quant funds, securities dealers, "market-data vendors, sportsbooks and sports-statistics providers", with 5 to 500 employees and no country limit.
This is a different thing from selling data to hedge funds. There, an operating company sells signals funds can trade on, as the guide to selling company data to hedge funds explains. Here the trading firm is the seller, and the buyer wants examples of market structure and engineering for model training.
What makes trading data and code worth more or less, and why does exclusivity matter?
The listing says its range "Depends on history and exclusivity". History means years of unbroken, documented series. Exclusivity means whether anyone else can license the same thing.
A buyer pays more for a set no competitor can get, and a firm gives up more by agreeing to that. The trade-off is covered in exclusive vs non-exclusive licences.
| Raises value | Lowers value |
|---|---|
| Many years of continuous history | Short or gappy series |
| Data only the firm could have recorded | Data a vendor already sells |
| Documented schema, clock source and known gaps | Raw captures with no documentation |
| Code with full commit history and tests | A snapshot with history removed |
| No third-party licence in the code or data | Vendor libraries and licensed data sets embedded |
| Never licensed before | Already licensed to others |
What about client information and record-keeping rules?
Client data stays out unless the privacy rules are met. Under Regulation S-P a covered firm "may not, directly or through any affiliate, disclose any nonpublic personal information about a consumer to a nonaffiliated third party unless" it has given the required notices and a chance to opt out (17 CFR 248.10). Institutional clients are protected by contract instead: check side letters, prime services agreements and non-disclosure agreements.
Licensing a copy does not change what a firm must keep. Registered securities firms "must preserve for a period of not less than 6 years, the first two years in an easily accessible place" some records, and others for "not less than three years" (17 CFR 240.17a-4). Investment advisers must keep required books and records "in an easily accessible place for a period of not less than five years from the end of the fiscal year during which the last entry was made" (17 CFR 275.204-2).
A fund that is winding down still owes those retention duties, and it can still license what it owns if someone has authority to sign. The steps are in the hedge fund wind-down guide. This page is general information, not legal advice.
Example: a trading firm lists what it actually owns
A fictional proprietary trading firm with 40 staff lists its sets and the source of each.
| Set | History | Source | Decision |
|---|---|---|---|
| Full-depth order-book captures, six venues | 9 years | Venue feeds under data agreements | Excluded: the venue data agreements govern redistribution and training use |
| Own orders, fills and cancels | 11 years | The firm's own activity | Candidate, after counterparty fields are removed |
| Intraday risk and inventory snapshots | 8 years | Internal systems | Candidate |
| Order management system and FIX and ITCH handlers | 10 years of commits | Written in-house; two vendor libraries | Candidate once vendor code is carved out |
| Research notebooks | 6 years | In-house, but embed a vendor's fundamentals data | Hold: strip embedded vendor data first |
Fictional example. The largest file on disk is the one that cannot be licensed. The firm's own activity and code are the asset.
What a licence usually allows and how long it takes
A licence for time series or code usually lets the buyer train and test its own models on the material under confidentiality. It usually bars resale, publication and use of the material to trade. Firms often add a delay so that recent data is held back.
Ask about exclusivity, the embargo on recent periods, where the data will be stored, and what happens to the copies at the end of the term. There is no standard timetable. The order is usually the same: a written description of the archive, a small sample, a rights review, agreed terms, then delivery. The rights review sets the pace. For a trading firm the slow step is reading data agreements and scanning code for third-party licences.
Common mistakes trading firms make
Nearly all of them are about ownership.
- Treating recorded exchange feeds as proprietary because the firm stored them.
- Calling a derived series proprietary without listing its inputs.
- Leaving client, counterparty or prime services identifiers in fills and positions.
- Sharing a repository with venue credentials, session configuration or live parameters in it.
- Skipping the licence scan on vendor libraries and open-source dependencies.
- Agreeing to exclusivity without pricing what it rules out.
- Deleting systems during a wind-down before retention duties and export are dealt with.
Trading data and code as a data asset
One published company range for hedge funds and quant firms runs from $25,000 to $1.15 million (one buyer's public listings, read 15 November 2026). Read it as a published ceiling, not as a typical outcome. The listing says the figure "Depends on history and exclusivity", and sets built on exchange feeds do not qualify at all.
Start with a source list for every series and a licence scan of the code. Payment and transaction records are a different asset, covered in the transaction data guide. Origin Data Partners is an independent data partner that helps companies get their data bought. A buyer may pay Origin when a deal closes.
Does your company hold records like these? See what your data is worth.
Common questions
Can a trading firm license its market data for AI training?
Yes, for time series it owns and has the right to license. A public listing read on 15 November 2026 asks for multi-year proprietary time series or trading infrastructure code. Redistributed exchange feeds and public data are not a fit.
Does exchange data a firm recorded count as its own?
Usually not. The data is licensed from the venue under a data agreement, and exchange policies restrict redistribution and treat model training as its own category of use. A firm's own orders, fills and internal risk data are the stronger candidates.
Is trading infrastructure code worth anything after a strategy is retired?
It can be. The listing names order management systems, FIX, ITCH and OUCH handlers, low-latency engines and backtesters as code that adds value, and a firm that owns such code qualifies without data. A third-party licence scan comes first.
How does exclusivity change what a trading data set is worth?
The listing says its range depends on history and exclusivity. A buyer pays more for a set no one else can license, and the firm gives up the chance to license it again. Price the restriction before agreeing to it.
Can a fund that is winding down license its data and code?
Yes, if it owns them and someone has authority to sign. Record retention rules, such as the five-year rule for investment advisers in 17 CFR 275.204-2, still apply, so systems are exported and preserved before they are switched off.
Can a sportsbook license its odds history?
The listing names sportsbooks and sports-statistics providers and lists sports odds and outcome histories. The lines the book set are its own work. Inputs taken from a league or official data supplier are governed by that supply contract.
Sources and further reading
- Cboe: Market Data Policies (PDF)
- FIX Trading Community: the FIX Protocol
- 18 U.S.C. 1839: definitions (trade secret)
- 17 CFR 248.10: limits on disclosure of nonpublic personal information (Regulation S-P)
- 17 CFR 240.17a-4: record preservation for registered securities firms
- 17 CFR 275.204-2: books and records to be maintained by investment advisers
