CryptoDataDownload – Bitstamp BTC/USD Hourly OHLCV
- Rows
- 74,503
- Columns
- 9
- Data range
- 2018-05-15T06:00:00 to 2026-11-22
- Auto-refresh
- Re-checks its source every 7 days
- Last refreshed with new data
- 2026-11-22
Hourly OHLCV data for BTC/USD on Bitstamp. Fields: Unix Timestamp, Date, Symbol, Open, High, Low, Close, Volume BTC, Volume USD. Multi-year history at 1-hour granularity. Free under CC BY-NC-SA 4.0. Suitable for academic research and backtesting.
AI analysis
Dataset Analysis: Bitstamp BTC/USD Hourly OHLCV
1. Dataset Overview and Correlation Value
This dataset, sourced from [CryptoDataDownload](https://www.cryptodatadownload.com/cdd/BitstampBTCUSD1h.csv), provides a rich, high-resolution historical record of Bitcoin price action on the Bitstamp exchange, spanning from May 2018 through June 2026 — over eight years of hourly granularity. The publisher is a well-known aggregator of free cryptocurrency market data, and the dataset's coverage through mid-2026 suggests it is actively maintained and refreshed via API pull. With 70,927 hourly candles across a period that includes the 2020–2021 bull run, the 2022 bear market, and subsequent recovery cycles, this dataset is exceptionally well-suited for time-series correlation studies, particularly when paired with macroeconomic, on-chain, or sentiment data at comparable or coarser time resolutions.
---
2. Data Quality Observations
The dataset presents an unusually clean profile for financial market data. Total null cells across all 9 columns is zero, and every column reports distinct=70,927 for the date and unix fields, confirming no duplicate timestamps exist — each hourly candle is uniquely represented. The symbol column carries exactly 1 distinct value (presumably "BTCUSD"), making it informationally inert but useful as a schema integrity marker. No type mismatches are reported. The one significant data quality concern lies in the Volume BTC and Volume USD columns, which both show a minimum of 0, suggesting the presence of zero-volume candles — likely representing periods of exchange downtime, data gaps filled with placeholders, or extremely illiquid hours. This warrants filtering before any volume-based analysis. Duplicate row counts are noted as pending Phase D recomputation against Parquet, but the timestamp distinctness effectively rules out meaningful duplication.
---
3. Key Column Distributions
The price columns (Open, High, Low, Close) are tightly co-distributed, as expected for OHLCV data. Close price has a mean of ~$40,596 but a median of only ~$30,476, with a rightward skew of 0.76, reflecting the asymmetric bull-run price spikes — particularly the late-2021 peak visible in the max of ~$126,111. The interquartile range spans roughly $10,173 (Q1) to $63,205 (Q3), a spread of over $53,000, underscoring Bitcoin's historical volatility across this period.
The Volume BTC column is far more revealing of data anomalies. Its mean of ~756,121 BTC per hour versus a median of just ~116 BTC per hour signals extreme right skew (skew = 7.9) and 12,113 flagged outliers — roughly 17% of all rows. The maximum of ~68.6 million BTC per hour is implausible given Bitcoin's total circulating supply, strongly suggesting either data errors, unit inconsistencies in early records, or exchange-reported anomalies. Volume USD similarly shows skew of 5.87 with 6,043 outliers, and a Q1 of just $665 versus a Q3 of ~$4.74 million, hinting at structural data heterogeneity across the multi-year span — possibly reflecting Bitstamp's growth in trading activity over time rather than pure noise.
---
4. Recommended Join Key Columns
The date column is the primary recommended join key, offering 70,927 distinct ISO timestamp values at hourly granularity with zero nulls — ideal for time-series alignment. The unix timestamp (range: ~1.526×10⁹ to ~1.782×10⁹) serves as a reliable numeric join key for systems that prefer epoch integers, and its near-zero skew (0.00114) confirms uniform temporal sampling with no clustering artifacts. For cross-dataset joins at lower resolution (daily, weekly), the date column can be truncated. When joining with on-chain or macroeconomic datasets that report at daily frequency, aggregating OHLCV to daily OHLCV (using open of first candle, close of last, volume sum) before joining is strongly recommended to avoid fan-out multiplication errors.
---
5. Suggested Paired Datasets for Correlation Discovery
| Suggested Dataset | Rationale | |---|---| | Bitcoin On-Chain Metrics (e.g., Glassnode, CryptoQuant) — active addresses, hash rate, exchange inflows | Direct causal candidates for price and volume movements; testable at daily granularity | | Google Trends — "Bitcoin" search interest | Retail sentiment proxy; known to lead or lag price during bubble cycles | | Federal Reserve / FRED — US CPI, Interest Rates, M2 Money Supply | Tests macro-inflation narrative; correlations likely strongest at monthly aggregation | | S&P 500 / Gold / DXY Hourly OHLCV | Risk-asset correlation analysis; particularly interesting post-2022 when BTC-equity correlation spiked | | Crypto Fear & Greed Index (Alternative.me — daily) | Sentiment-to-price lag/lead analysis; widely cited in academic crypto literature | | Twitter/X BTC Mention Volume or Sentiment Scores | Social signal correlation; volume outliers in this dataset may align with viral events |
The volume outlier concentration (~17% of rows for BTC volume) makes on-chain exchange flow data a particularly valuable pairing to validate or explain anomalous trading periods before building any predictive models.
Columns
- date (date)
- unix (integer)
- symbol (string)
- open (decimal)
- high (decimal)
- low (decimal)
- close (decimal)
- Volume BTC (decimal)
- Volume USD (decimal)