lynx1231/historical-major-usd-fx-forwards-sample

Historical Major USD FX Forwards Sample

A free evaluation sample of historical FX-forward data for selected major USD-linked currency pairs across multiple forward tenors.

Full historical FX-forward catalog, broader pair and tenor coverage, downloadable datasets, and pricing:
https://futuresforexandsomeindexes.com/

This repository is a free evaluation sample intended for schema inspection, data-quality evaluation, integration testing, and quantitative research prototyping. It is not the complete commercial dataset.

Sample coverage

This release contains 4 currency pairs, 12 pair/tenor configurations, 12 distinct monthly partition codes, and 3,726,475 observations across daily and minute frequencies.

Currency pair	Base / cross	Tenors	Daily source partitions	Minute source partitions
AUDUSD	AUD/USD	1M, 3M, 1Y	36	36
EURUSD	EUR/USD	1M, 3M, 1Y	36	36
GBPUSD	GBP/USD	1M, 3M, 1Y	36	36
USDJPY	USD/JPY	3M, 6M, 1Y	36	36

The sample manifest determines the exact tenor and monthly partitions included. The complete catalog may contain substantially broader pair, tenor, and historical coverage.

Package contents

• data/daily/*.parquet — normalized daily FX-forward observations.
• data/minute/*.parquet — normalized minute-level FX-forward observations.
• catalog.csv / catalog.parquet — one row per sampled pair/tenor/frequency/month source file with observation counts, timestamp bounds, output shard, and SHA-256.
• metadata.json — release-level coverage and file statistics.
• manifest.json — sanitized public sample manifest; internal tenor and pair IDs are excluded.
• schema.json — column definitions plus timestamp and volume semantics.
• checksums.sha256 — package integrity hashes.
• dataset-metadata.json — Kaggle publication metadata.
• examples/ — Polars, pandas, and DuckDB loading examples.

Data format

The release uses Apache Parquet with ZSTD compression.

Daily and minute files share the same normalized columns:

timestamp, open, high, low, close, volume, pair_name, tenor, dataset, base, cross, file_code

Daily observations in this release: 3,096
Minute observations in this release: 3,723,379

file_code preserves the monthly source partition label, such as 01-25.

Forward tenor

tenor identifies the forward horizon represented by the observation, such as 1M, 3M, 6M, or 1Y. Not every pair in the evaluation sample necessarily uses the same tenor set.

Volume

The current FX-forward staging source contains OHLC values but does not populate volume. The volume field is therefore retained as nullable for schema stability and future compatibility; consumers should not treat null volume as zero trading activity.

Timestamp semantics

timestamp is the original source timestamp integer and is preserved unchanged. The public release builder does not perform timezone conversion.

Quick start

Polars

import polars as pl

minute = pl.scan_parquet("data/minute/*.parquet")

sample = (
    minute
    .filter(
        (pl.col("pair_name") == "EURUSD")
        & (pl.col("tenor") == "1M")
    )
    .collect()
)

print(sample.head())

pandas

from pathlib import Path
import pandas as pd

files = sorted(
    Path("data/minute").glob("*.parquet")
)

df = pd.concat(
    (pd.read_parquet(path) for path in files),
    ignore_index=True,
)

print(df.head())

DuckDB

SELECT
    pair_name,
    tenor,
    COUNT(*) AS observations
FROM read_parquet('data/minute/*.parquet')
GROUP BY pair_name, tenor
ORDER BY pair_name, tenor;

Typical uses

• Forward-curve and tenor-structure analysis across selected major USD currency pairs.
• Daily versus intraday FX-forward research and market-data pipeline development.
• Cross-tenor feature engineering, relative-value research, and integration testing.
• Schema and data-quality evaluation before obtaining broader historical pair and tenor coverage.

Integrity and reproducibility

Every Parquet shard is referenced in the catalog and the package includes checksums.sha256. Before release, the builder validates required columns, row-level manifest identifiers, required non-null fields, basic OHLC bounds, tenor-source uniqueness, pair-source consistency, and monthly partition uniqueness.

Internal source identifiers used by the staging pipeline are intentionally excluded from public release metadata.

Complete dataset

This package is intentionally limited to a small evaluation subset. For the complete FX-forward catalog, broader pair and tenor coverage, availability, samples, and pricing, visit:

https://futuresforexandsomeindexes.com/

Release information

• Release version: 2026-09-18
• Generated: 2026-09-18T16:24:55.240782+00:00
• Dataset family: Major USD FX Forwards
• Compression: zstd
• License metadata: other