GUIDE · XAU/USD · TIMEFRAMES · CATALOG VERIFIED 2026-10-08

XAU/USD timeframes: resample 1-minute (M1) data to M5, H1, H4 and daily

Any XAU/USD timeframe can be built from 1-minute bars with one pandas resample call; the only real choice is where a daily bar starts. AI Options Data Store sells XAU/USD spot M1 OHLC from 15 Mar 2009 to 2 Oct 2026 as Parquet in UTC, so M5, M15, H1, H4, daily and weekly bars are a few lines away.

Get the XAU/USD pack XAU/USD historical data M1 data and CSV Free sample (no email)

Download the sample → check the columns

Dates on this page are read from the live file catalog, not typed by hand. Full matrix: /coverage.

Timeframe to pandas rule

Timeframe	pandas rule	Note
M5 / M15 / M30	'5min' / '15min' / '30min'	Bars labelled by their start
H1 / H4	'1h' / '4h'	H4 buckets start at 00:00 UTC
D1 (UTC day)	'1D'	Day = 00:00–24:00 UTC
D1 (17:00 New York close)	'1D' on a shifted index	Common for spot charts; see code
W1	'W-FRI'	Week ending Friday

Intraday bars (M5 to H4)

Aggregate open as first, high as max, low as min and close as last. dropna() removes empty buckets (weekends, holidays); nothing is forward-filled.

import pandas as pd

df = pd.read_parquet('XAUUSD_SPOT_1min_2025-01-01_2025-12-31.parquet').set_index('datetime')
AGG = {'open': 'first', 'high': 'max', 'low': 'min', 'close': 'last'}

m15 = df.resample('15min').agg(AGG).dropna()
h1  = df.resample('1h').agg(AGG).dropna()
h4  = df.resample('4h').agg(AGG).dropna()

Daily bars with a 17:00 New York close

Shift New York local time by seven hours so 17:00 becomes midnight, then resample by day. Because the shift is applied in local time, daylight-saving changes are handled.

ny = df.copy()
ny.index = df.index.tz_convert('America/New_York') + pd.Timedelta(hours=7)
d1 = ny.resample('1D').agg(AGG).dropna()     # each bar ends 17:00 New York
w1 = ny.resample('W-FRI').agg(AGG).dropna()
d1.to_csv('xauusd_d1.csv')                   # CSV if needed

Same thing in DuckDB

DuckDB reads the yearly Parquet files with a glob and aggregates without loading everything into pandas.

import duckdb

duckdb.sql("SET TimeZone = 'UTC'")
h1 = duckdb.sql("""
  SELECT time_bucket(INTERVAL '1 hour', datetime) AS ts,
         arg_min(open, datetime) AS open, max(high) AS high,
         min(low) AS low, arg_max(close, datetime) AS close
  FROM 'XAUUSD_SPOT_1min_*.parquet'
  GROUP BY ts ORDER BY ts
""").df()

Exactly which fields are in the files

XAU/USD 1-minute files

Column	Type	Meaning
datetime	timestamp (UTC)	Minute start, timezone-aware UTC
open / high / low / close	float	Bid-side XAU/USD quote OHLC, US dollars per troy ounce

Not included: ask-side prices, futures, options, implied volatility or Greeks. 1-minute bars, not tick data.

Format: Parquet (XAUUSD_SPOT_1min_….parquet), UTC. The free September 2026 sample is enough to test every snippet.

FAQ

Can I get XAU/USD hourly or daily data from 1-minute data?

Yes. Resample the 1-minute bars (first/max/min/last). AI Options Data Store sells XAU/USD M1 OHLC from 15 Mar 2009 to 2 Oct 2026 as Parquet in UTC, so every higher timeframe can be rebuilt exactly.

Which daily close should I use?

Either the UTC day or a 17:00 New York close, which many spot charts use. The guide shows both; pick one and keep it consistent across tests.

Why do my H4 bars not match my charting platform?

Platforms differ in where the day and the 4-hour buckets start. Apply the same offset (for example the New York shift) before resampling.

Are empty weekend bars created?

pandas creates empty buckets for the closed hours; dropna() removes them. No prices are forward-filled.

Can I export the result to CSV?

Yes, df.to_csv() on any resampled frame.

Get the data

Free sample · Schema · Ask the AI for your data

Get the XAU/USD pack XAU/USD historical data M1 data and CSV Free sample