wehub-resource-sync
89 lines
2.7 KiB
Python
89 lines
2.7 KiB
Python
"""Fama-French and AQR factor loaders."""
|
|
|
|
from pathlib import Path
|
|
from typing import Literal
|
|
|
|
import polars as pl
|
|
|
|
from data.exceptions import DataNotFoundError
|
|
from utils import ML4T_DATA_PATH
|
|
|
|
|
|
def load_ff_factors(
|
|
dataset: Literal["ff3", "ff5", "mom"] = "ff5",
|
|
frequency: Literal["daily", "monthly"] = "monthly",
|
|
start_date: str | None = None,
|
|
end_date: str | None = None,
|
|
) -> pl.DataFrame:
|
|
"""Load Fama-French factor returns.
|
|
|
|
Args:
|
|
dataset: "ff3" (3-factor), "ff5" (5-factor model), or "mom" (momentum factor)
|
|
frequency: "daily" or "monthly"
|
|
start_date: Optional start date filter (YYYY-MM-DD format)
|
|
end_date: Optional end date filter (YYYY-MM-DD format)
|
|
|
|
Returns:
|
|
DataFrame with factor returns and a timestamp/date column.
|
|
"""
|
|
filename = f"{dataset}_{frequency}.parquet"
|
|
path = ML4T_DATA_PATH / "factors" / "fama-french" / filename
|
|
if not path.exists():
|
|
raise DataNotFoundError(
|
|
dataset_name="Fama-French Factors",
|
|
path=path,
|
|
download_script="data/factors/ff_download.py",
|
|
readme="data/factors/README.md",
|
|
)
|
|
|
|
df = pl.read_parquet(path)
|
|
|
|
# Normalize to canonical schema
|
|
if "date" in df.columns and "timestamp" not in df.columns:
|
|
df = df.rename({"date": "timestamp"})
|
|
if "timestamp" in df.columns and df["timestamp"].dtype != pl.Date:
|
|
df = df.with_columns(pl.col("timestamp").cast(pl.Date))
|
|
|
|
if "timestamp" in df.columns:
|
|
if start_date:
|
|
df = df.filter(pl.col("timestamp") >= pl.lit(start_date).str.to_date())
|
|
if end_date:
|
|
df = df.filter(pl.col("timestamp") <= pl.lit(end_date).str.to_date())
|
|
|
|
return df
|
|
|
|
|
|
def load_aqr_factors(
|
|
dataset: Literal["qmj", "bab", "hml_devil", "vme"] = "qmj",
|
|
) -> pl.DataFrame:
|
|
"""Load AQR factor returns.
|
|
|
|
Args:
|
|
dataset: Factor dataset
|
|
- "qmj": Quality Minus Junk
|
|
- "bab": Betting Against Beta
|
|
- "hml_devil": HML Devil (quality-adjusted value)
|
|
- "vme": Value-Momentum Everywhere
|
|
|
|
Returns:
|
|
DataFrame with factor returns across geographies
|
|
"""
|
|
# Files have inconsistent naming - use lookup
|
|
file_map = {
|
|
"qmj": "qmj_factors.parquet",
|
|
"bab": "bab_factors.parquet",
|
|
"hml_devil": "hml_devil.parquet",
|
|
"vme": "vme_factors.parquet",
|
|
}
|
|
filename = file_map[dataset]
|
|
path = ML4T_DATA_PATH / "factors" / "aqr" / filename
|
|
if not path.exists():
|
|
raise DataNotFoundError(
|
|
dataset_name="AQR Research Factors",
|
|
path=path,
|
|
download_script="data/factors/aqr_download.py",
|
|
readme="data/factors/README.md",
|
|
)
|
|
|
|
return pl.read_parquet(path)
|