File size: 2,565 Bytes
05707f0 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 | """Size and power of the out-of-sample tests on known-truth synthetic panels.
Writes reports/validation/size_power.csv and .md. This is software validation:
every number here comes from simulated data with a known data-generating process.
python scripts/size_power.py --reps 40
"""
from __future__ import annotations
import argparse
import sys
from pathlib import Path
import pandas as pd
sys.path.insert(0, str(Path(__file__).resolve().parents[1] / "src"))
from emft.backtest import run_backtest # noqa: E402
from emft.data import synthetic_panel # noqa: E402
from emft.evaluate import r2_table # noqa: E402
from emft.features import build_features, model_ready # noqa: E402
CRIT = 1.645 # one-sided 5%
def one(signal_sd: float, seed: int) -> dict:
panel, _ = synthetic_panel(n_dm=6, n_em=6, n_factors=12, n_months=300, signal_sd=signal_sd, seed=seed)
d = model_ready(build_features(panel))
t = r2_table(run_backtest(d, models=("ols",), scopes=("own",), oos_start=2005)).iloc[0]
return {"signal_sd": signal_sd, "seed": seed, **{k: t[k] for k in
("r2_vs_hist_pct", "cw_t_hist", "r2_vs_shrink_pct", "cw_t_shrink")}}
def main() -> None:
ap = argparse.ArgumentParser()
ap.add_argument("--reps", type=int, default=40)
ap.add_argument("--out", default="reports/validation")
a = ap.parse_args()
rows = [one(s, seed) for s in (0.0, 0.004, 0.008) for seed in range(100, 100 + a.reps)]
df = pd.DataFrame(rows)
summ = df.groupby("signal_sd").agg(
reps=("seed", "size"),
mean_r2_vs_hist=("r2_vs_hist_pct", "mean"),
reject_vs_hist=("cw_t_hist", lambda t: (t > CRIT).mean()),
mean_r2_vs_shrink=("r2_vs_shrink_pct", "mean"),
reject_vs_shrink=("cw_t_shrink", lambda t: (t > CRIT).mean()),
).reset_index()
out = Path(a.out)
out.mkdir(parents=True, exist_ok=True)
df.to_csv(out / "size_power_draws.csv", index=False)
summ.to_csv(out / "size_power.csv", index=False)
lines = [
"# Size and power on synthetic panels (software validation, not a finding)",
"",
"6 DM + 6 EM countries, 12 factors, 300 months, OOS from 2005, pooled OLS.",
"Rejection = Clark-West t > 1.645 (one-sided 5%). Row signal_sd = 0 is the null.",
"",
summ.round(3).to_markdown(index=False) if hasattr(summ, "to_markdown") else summ.round(3).to_string(index=False),
]
(out / "size_power.md").write_text("\n".join(lines) + "\n")
print(summ.round(3).to_string(index=False))
if __name__ == "__main__":
main()
|