Datasets

Manage the historical market data used for backtesting. Import your own CSV files or fetch data directly from supported providers.

Sample data

Augy comes with a pre-loaded Sample BTC-USD 1H dataset containing 2,160 hourly bars (approximately 90 days). This lets you start backtesting immediately without configuring any data provider.

Import CSV

Upload any OHLCV CSV file. The parser auto-detects common column name variations:

FieldAccepted column names
Timestamptimestamp, date, datetime, time
Openopen, o
Highhigh, h
Lowlow, l
Closeclose, c
Volumevolume, vol, v (optional, defaults to 0)

After selecting a file, you'll see a preview showing the parsed bar count, date range, and any parsing warnings. Then specify a name, symbol, and timeframe for the dataset.

Fetch from data providers

Requires API keys configured in Settings.

Twelve Data

Fetch data for stocks, forex, and crypto symbols.

  • Stocks: AAPL, MSFT, TSLA
  • Forex: EUR/USD, GBP/USD
  • Crypto: BTC/USD, ETH/USD

Free tier: 8 API calls/minute, up to 5,000 bars per request. Augy automatically paginates for larger date ranges with rate-limit pauses.

Alpaca

Fetch data for US stocks and crypto.

  • Stocks: AAPL, MSFT, TSLA (SIP feed with split adjustment)
  • Crypto: BTC/USD, ETH/USD (free, no paid plan needed)

Up to 10,000 bars per page with automatic pagination.

Supported timeframes

1m5m15m30m1h2h4h1d1w

Select the timeframe when importing CSV or fetching from a provider. The timeframe is stored with the dataset and displayed on the Datasets page.

How data is stored

All datasets are cached locally in SQLite. When you run a backtest, Augy exports the selected dataset to a temporary CSV file that the Python backtesting engine reads. This means:

  • Data is fetched once and reused across multiple backtests.
  • No network calls are needed when re-running a backtest.
  • Deleting a dataset removes all its data from SQLite.