Wiki Prices Provider¶
Provider: WikiPricesProvider
Data Source: Local Parquet file
API Key: Not required
Free Tier: Free (local file)
Overview¶
Wiki Prices provides historical US equity data from 1962-2018, including delisted companies for survivorship-bias-free research.
Best For: Long-term backtesting, survivorship-bias-free research
Quick Start¶
from ml4t.data.providers import WikiPricesProvider
# Point to local Parquet file
provider = WikiPricesProvider(parquet_path="~/data/wiki_prices.parquet")
# Fetch historical data
df = provider.fetch_ohlcv("AAPL", "1990-01-01", "2018-03-27")
provider.close()
Dataset Details¶
| Attribute | Value |
|---|---|
| Coverage | 3,199 US stocks |
| Rows | 15.4 million |
| Period | 1962-2018 |
| Size | ~631 MB (Parquet) |
| Includes Delisted | Yes |
Fallback Pattern¶
Combine with Yahoo Finance for complete history:
from ml4t.data.providers import WikiPricesProvider, YahooFinanceProvider
import polars as pl
# Pre-2018: Wiki Prices
wiki = WikiPricesProvider(parquet_path="~/data/wiki_prices.parquet")
historical = wiki.fetch_ohlcv("AAPL", "1990-01-01", "2018-03-27")
# Post-2018: Yahoo Finance
yahoo = YahooFinanceProvider()
recent = yahoo.fetch_ohlcv("AAPL", "2018-03-28", "2024-12-01")
# Combine
combined = pl.concat([historical, recent]).sort("timestamp")
print(f"Total: {len(combined)} rows from 1990-2024")
Data Source¶
The Wiki Prices dataset was originally from Quandl's WIKI database before Quandl discontinued the free tier.