Direct answer
Forex data comes from places that observe or compute market information—such as trading venues and electronic execution systems that maintain order books, plus brokers and market-data vendors that aggregate, normalize, and distribute that information through data feeds.
How it works (common data sources)
Forex “data” usually means one or more of these: prices (quotes), executed trades, volumes, and sometimes derived fields such as spreads or volatility. The raw building blocks typically come from:
- Trading systems (primary observation): Electronic platforms record events like bid/ask quotes and executed trades. Data produced here reflects what those systems saw.
- Brokers (compiled distribution): Brokers can provide their clients with market data sourced from one or more upstream liquidity providers. They may also combine streams, apply symbol mappings, and publish a consistent feed.
- Market-data vendors (aggregation and normalization): Vendors ingest data from multiple venues and brokers, then standardize formats, align instruments, and may fill gaps depending on their methodology.
- Derivations and reference series (computed from inputs): Some datasets are not direct observations; they are calculated from other inputs (for example, selecting representative prices or producing averages). These series are useful but depend on the rule set used to compute them.
Example checks to understand source differences
Even without knowing a provider’s internal process, you can compare how “the same” pair appears across sources:
- Quotes vs. trades: One feed may show many quote updates but fewer executed trades, changing the apparent activity.
- Timing consistency: If two sources show the same price moves but with different delays or time stamps, they are capturing and timestamping events differently.
- Symbol mapping: A feed might represent instrument identifiers differently (for example, contract naming), which can lead to mismatches when comparing.
- Data completeness: Some feeds may smooth, filter, or omit certain updates; this can affect spreads, perceived liquidity, and apparent volatility.
Limitations and what to verify
Forex data is not a single universal stream. Differences in upstream observation, aggregation rules, symbol mapping, and timestamp handling can make datasets disagree.
To independently verify data quality, compare at least two independent sources for the same instrument and time window, and check whether discrepancies are systematic (for example, consistent time offsets) or random (varying widely). Also assume that “derived” fields reflect the vendor’s computation rules, not a direct market measurement. No future performance or guaranteed accuracy can be inferred from historical data alone.