Datasets

skfolio comes with three native datasets available via:

• load_sp500_dataset

• load_sp500_index

• load_factors_dataset

Larger datasets are downloaded from the GitHub repo and cached locally to a data directory. They are available via:

• load_ftse100_dataset

• load_nasdaq_dataset

• load_sp500_implied_vol_dataset

By default the data directory is set to a folder named “skfolio_data” in the user home folder. Alternatively, it can be set by the SKFOLIO_DATA environment variable. If the folder does not already exist, it is automatically created.

For characteristics-based factor models, make_synthetic_characteristics generates a synthetic AssetPanel with the market and fundamental fields used by the default descriptors and CharacteristicsFactorModel. Use it for examples, tests, and local prototyping when you do not have a point-in-time fundamentals feed.

Caution

This dataset is provided solely for testing and example purposes. It is a stale dataset and does not reflect current or accurate market prices. It is not intended for investment, trading, or commercial use and should not be relied upon as authoritative market data.

Example:

Loading the S&P 500 dataset, which contains daily adjusted closing prices for 20 selected constituents of the S&P 500 Index, covering the period from 1990-01-02 to 2022-12-28:

from skfolio.datasets import load_sp500_dataset

prices = load_sp500_dataset()
print(prices.head())

Generating a synthetic characteristics panel:

from skfolio.datasets import make_synthetic_characteristics

panel = make_synthetic_characteristics(n_assets=200, n_observations=1000)
print(panel.n_assets, panel.n_observations)