# Provenance stamped 2026-10-07 (roadmap#943): written by an API Evangelist sweep (dated 2026-06-13 by the file itself), not
# harvested from the provider. `method: generated` so the economics campaign may replace it with a
# dated, verbatim transcription; every figure below is unverified until it does.
method: generated
generated: '2026-06-13'
specification: API Commons Rate Limits
specificationVersion: '0.1'
schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/RateLimits
provider: OANDA
providerId: oanda
created: '2026-06-13'
modified: '2026-06-13'
reconciled: true
tags:
  - Rate Limiting
  - Forex
  - Trading
description: >-
  OANDA v20 REST API rate limits apply per authenticated user. New streaming connections
  are throttled to 2 per second; persistent HTTP connections support up to 100 requests
  per second. Aggregate streaming connections are capped at 20 across all sessions.
sources:
  - https://developer.oanda.com/rest-live-v20/best-practices/
  - https://developer.oanda.com/rest-live-v20/api-comparison/
responseCodes:
  throttled: 429
limits:
  - name: Per-user request limit
    scope: user
    metric: requests
    limit: 30
    timeFrame: rolling
    description: Maximum number of concurrent in-flight requests per authenticated user.
  - name: New connection rate
    scope: global
    metric: connections_per_second
    limit: 2
    timeFrame: second
    description: New TCP/HTTP connections are throttled to 2 per second per client.
  - name: Persistent connection throughput
    scope: connection
    metric: requests_per_second
    limit: 100
    timeFrame: second
    description: >-
      Existing persistent HTTP connections can sustain up to 100 requests per second.
      HTTP 1.1 keep-alive is enabled by default; HTTP 1.0 clients should send
      Connection: Keep-Alive header.
  - name: Aggregate streaming connections
    scope: account
    metric: connections
    limit: 20
    timeFrame: session
    description: >-
      Total concurrent streaming connections (pricing stream and transaction stream)
      across all sessions for a single account.
  - name: Historical candle records per page
    scope: request
    metric: records
    limit: 5000
    timeFrame: request
    description: Maximum number of candlestick records returned per candle history request.
  - name: FIX API default message rate
    scope: session
    metric: messages_per_second
    limit: 100
    timeFrame: second
    description: Default FIX API message rate limit (institutional accounts only; configurable).
policies:
  - name: Persistent connections preferred
    description: >-
      Use HTTP persistent (keep-alive) connections to reduce per-request latency and
      avoid triggering the new connection rate throttle.
  - name: Account state sync pattern
    description: >-
      Initialise with a full account snapshot via GET /v3/accounts/{accountID}, then
      poll GET /v3/accounts/{accountID}/changes with the last TransactionID to receive
      only incremental updates.
  - name: Streaming connection management
    description: >-
      Open streaming connections sparingly; reconnect with exponential backoff on
      disconnect. Do not exceed 20 aggregate streaming connections per account.
  - name: Backoff on errors
    description: >-
      Implement exponential backoff when receiving rate limit or server error responses
      to avoid sustained throttling.