PortfolioSort

Repository to perform portfolio sorts in empirical asset pricing. This function performs cross-sectional sort of assets, conditional on a characteristic exposure. The function returns (weighted) portfolio returns, test statistics, portfolios populations or the portfolios time-series.

Additionally, overhead is reduced when speedups are activated, which makes this function applicable for repetitive usage. Bottlenecks are furthermore just-in-time compiled using Numba.

Example:

from PortSort import PortfolioSort


# Generate or import some time series data
returns = <...>
characteristic = <...>
market_value = <...>

# Perform portfolio sorts on a charactistic, using 5 portfolios.
result = PortfolioSort.single_sort(df_char=characteristic,
                                   df_ret=returns,
                                   df_mcap=market_value,
                                   quantiles=[.2, .4, .6, .8, 1])