Free Bitcoin OHLCV Dataset — 6 Timeframes, Ready for Backtesting

If you're building a trading bot or backtesting a strategy, raw tick data is usually more than you need. What most developers actually want is clean OHLCV candles across multiple timeframes — ready to load and go.

We just published a free 1-week sample of our Bitcoin OHLCV Pack: six timeframes (1m, 5m, 15m, 1h, 4h, 1d), aggregated from tick-level trade data, in Apache Parquet format.

What's in the sample

• Period: August 5-12, 2024 (one week, includes real volatility)
• Base resolution: 11,520 one-minute candles
• Derived timeframes: 5m, 15m, 1h, 4h, 1d — all aggregated from the same 1-minute base, with verified scaling ratios (5m ~ 1m/5, 15m ~ 1m/15, etc.)
• Format: Apache Parquet, ZSTD compressed
• Zero look-ahead bias: every candle's open/close is computed with explicit chronological ordering, verified before packaging

Quick start

import pandas as pd

df = pd.read_parquet("BTC_OHLCV_1m_SAMPLE.parquet")
print(df.head())
print(f"{len(df):,} candles from {df['candle_time'].min()} to {df['candle_time'].max()}")

Why OHLCV instead of raw ticks

Most bot developers don't need billions of individual trades — they need candles. Aggregating tick data into multiple timeframes correctly is more subtle than it looks: naive bucketing approaches can silently produce identical candle counts across different timeframes (a bug that's easy to miss if you don't cross-check the ratios). We validate every timeframe against its neighbors before anything ships.

Get the full dataset

The sample covers one week. The full Bitcoin OHLCV Pack covers 6.8 years (2020-2026) across the same six timeframes - $19, one-time purchase, no subscription.

Full dataset: https://jalvart.gumroad.com/l/ttrtkx (code GLITCH20 for 20% off)

More free samples and datasets: theglitchlist.com?utm_source=devto&utm_medium=referral&utm_campaign=free_sample

---

Have questions about the aggregation methodology or want a different timeframe combination? Drop a comment below.