Barchart: market-data feeds, APIs and historical delivery

Barchart supplies market-data infrastructure for firms building trading applications, research systems, charts and information services. Openfeed is its unified streaming and snapshot layer. These services can sit inside a broker’s technology stack without making Barchart the broker, execution venue or trading counterparty. The practical starting point is the data workflow and permitted audience.[2]

Coverage spans exchange-traded markets, OTC data and benchmarks. The documented inventory includes FX, equities, futures, options, crypto, indices, rates and physical-commodity data, including forward curves. Availability can be real time, delayed, end of day or next day. Depth is venue-dependent: the exchange list gives L2 futures examples including CME, ICE, Eurex and SGX. A broad catalogue is not an entitlement to every venue or data level.[3]

OnDemand’s reference catalogue is useful for separating a price-feed project from enrichment work. It lists corporate actions, financial statements, SEC filings, technical indicators and specialised futures/options services alongside quotes and history. Endpoint selection therefore affects the fields, update cadence and interpretation a broker must implement; a single “market data” line in a procurement specification leaves substantial scope unresolved.[28]

Openfeed uses Google Protocol Buffers. Its protocol files separate market messages, instrument definitions and client/server control, allowing generated encoders and decoders in supported languages. That division matters operationally: an application must process session and instrument information as well as prices. A decoder that reads trade fields alone is insufficient for maintaining a correctly interpreted live feed.[6]

Subscriptions can select individual symbols or a full exchange feed. Full-exchange delivery includes definitions and metadata as well as market updates and requires account configuration through Institutional Support. Barchart recommends Java, .NET or Go for a primary full-data pipe, with Python suited to analytics/ETL and JavaScript to dashboards; any supported SDK can handle a limited-symbol use case. These are workload recommendations, rather than a guarantee for a particular deployment.[9]

Openfeed sends instrument metadata at initialisation and does not repeat every descriptive field with each update. Consumers need local definitions, including price formatting. REST users require a separate metadata workflow and local reference tables. The REST documentation supports JSON/XML selection and currently labels its OpenAPI specification “coming soon”; the live interface documentation should be distinguished from an available machine-readable specification.[15][8]

The developer overview suggests streaming for universes above roughly 2,000 instruments and REST for smaller query-driven needs. This is design guidance, not a hard account limit. Continuous observation, query frequency, fan-out and application state can make streaming appropriate even with fewer symbols. The intended workload should be agreed during sizing.[1]

OnDemand offers custom packages for different business sizes, without a public price in the reviewed product overview. Openfeed’s access process uses a data consultant to establish the application, coverage, connectivity and licensing. The commercial enquiry should describe whether data is for internal use, client display, a subscription product, downloads or historical analysis.[26][4]

No enterprise fee schedule or executed customer agreement was established from the accessible official material. The signed offer therefore needs to settle recurring charges, any exchange charges, renewal and termination, and rights to retain or redistribute stored data. Website access terms are addressed separately below; they are not a substitute for an enterprise data licence.[29]

OnDemand requires HTTPS and an API key; its quick start says to keep the key out of browser-side code and public repositories. Market Replay similarly documents backend/job-runner use with credentials held in environment variables. Where an API carries credentials in request parameters, application and proxy logs need corresponding secret handling. This security design belongs in the integration, not only in a later production checklist.[27][20]

OnDemand’s overview describes multiple US AWS data centres and Equinix facilities for physical backup and low-latency requirements. That is a product-level infrastructure description, not a promise that a particular feed is hosted in a chosen jurisdiction or that all endpoints share the same recovery arrangement.[26]

Institutional Support handles Openfeed account and exchange configuration. For subscription problems, the FAQ asks for the username, endpoint, requested exchange/service type and complete subscription response. This distinguishes an entitlement issue from a connection failure. The public helpdesk supplies ticket submission and a knowledge base; response and restoration commitments need to be established for the selected enterprise service.[5][33]

Barchart’s operational guidance says the server sends a heartbeat every ten seconds. Monitor message timestamps, queues and processing lag as well as connection status: a connected consumer can still be stale. Overlapping logins with the same username may terminate both sessions, so reconnect logic should use controlled retries. If backlog becomes unusable, the guidance is to disconnect, clear stale work, reconnect and resubscribe, with historical delivery available for backfill. These are documented integration practices, not a measured uptime result.[18]

Historical-data troubleshooting starts with endpoint enablement, available date range and timezone. Tick record limits may be exceeded to include records sharing a matching second, so downstream code should not assume the requested maximum is an absolute payload cap. CSV layouts also vary by query. Logging the query type and parsing contract is more useful than treating every empty or unexpected response as an outage.[25]

A requested real-time subscription may be downgraded to delayed data if the account has only delayed permission; unavailable access can return INSUFFICIENT_PRIVILEGES. The application should inspect the granted response instead of assuming the request succeeded at the requested timeliness. Barchart’s user-settings interface exposes the exchanges and service types associated with a username, providing another configuration check.[5][14]

OnDemand’s overview expressly restricts the documented data to internal use and prohibits redistribution. It also asks customers to avoid personally identifiable information and use meaningless unique user identifiers. Client display, downloads and onward supply therefore need an expressly agreed licence rather than an assumption based on API access.[26]

The Websol website terms permit personal, noncommercial use and require prior written consent from Barchart and/or its data providers for reproduction or distribution. They describe website registration, cookies and personal information. Those provisions should not be read as an enterprise processing agreement or a commitment to export and erase a brokerage’s complete stored market-data estate. The customer agreement must define that service-specific responsibility.[30]

The public Websol information agreement names Barchart.com, Inc. / AgriCharts and describes information owned by or licensed to it. Barchart’s own publisher biography on Nasdaq identifies Chicago as its headquarters and dates its website origins to 1995.[29][31]

Barchart combines institutional data services with public market-information and media properties. A Barchart-issued announcement dated 11 June 2025 describes Plus500 sponsorship of its retail futures pages and related educational placements. That commercial arrangement illustrates the media side of the business. It is separate from the enterprise feed licence and does not make Barchart the clearing broker or confer its partner’s financial permissions.[32]

Operator

Barchart.com, Inc.

Service scope

Institutional real-time, delayed, snapshot and historical market data, with dataset-specific access and distribution permissions.

Customer types

Brokerages; Market makers; Financial institutions; Application developers; Research teams; Media companies

Delivery

API; Hosted

Pricing publication

Quote only

Platforms

Openfeed; OnDemand; Market Replay; Enterprise historical delivery

Pricing scope

Custom data packages and enterprise scope; no public institutional tariff established.

Implementation

Demo onboarding, entitled production credentials, instrument metadata and streaming or query integration.

Support

Institutional Support for Openfeed configuration plus the published helpdesk; purchased service terms define incident commitments.

Permissions and regulatory scope

Market-information technology and data licensing; no brokerage or execution permission is implied.

Provider metadata checked 20 Oct 2026Source 31Source 29Source 4Source 2Source 3Source 13Source 7Source 26Source 11Source 33Source 1Source 28

Minute history supplies OHLCV bars with a configurable interval, defaulting to one minute. The documentation uses Eastern Time for equities and Central Time for futures/FX; start boundaries are inclusive and end boundaries exclusive. Preserve the market timezone and interval with the exported dataset before aligning prices with a broker’s execution timestamps or a UTC reporting system.[22]

Market Replay separates trade ticks from bid/ask queries, with a dedicated quote route and query type. Historical quote observations should therefore be selected explicitly when analysing spread or execution context; a trade-only extract answers a different question. The one-day request window also needs to be preserved when assembling longer histories.[21]

Openfeed depth subscriptions build a local order book from an initial marketSnapshotDepthPriceLevel and subsequent add, modify and delete updates. Maintaining that state is the consumer’s responsibility. The documented price-level feed provides market information; it is not an order-entry or execution interface. This distinction matters when a broker combines displayed depth with a separate routing system.[19]

Corporate-event queries return split, dividend and earnings records. Omitting all event flags requests all available event types. Unknown symbols are ignored, and a query can have no row when no matching event exists. An empty result consequently needs symbol and date-range validation before it is interpreted as proof that no corporate action occurred.[24]