Signal construction

Form 4: For each company-day, we aggregate insider buy/sell transaction amounts and compute net buy ratio = (buy_$) / (buy_$ + sell_$), shrunk toward 0.5 with n/(n+κ) for small sample sizes (κ=10). 8-K: abnormal count vs trailing base rate. All signals are cross-sectionally ranked within date and sector.

Limitations & caveats

Insiders transact for non-informational reasons (10b5-1 plans, diversification, taxes). Form 4 has a 2-day filing window so observed_at can lag the transaction date. Institutional (13F) filings are quarterly.

Evaluation results

Rank IC (Spearman) with Newey-West standard errors, walk-forward with 21-day embargo. Multiple-testing correction: Benjamini-Hochberg at q=0.05 across all signal × horizon × cohort tests.

No results yet. Data collection is ongoing. Results appear once sufficient out-of-sample history has accumulated. See the power analysis for timeline estimates.

Data coverage

S&P 500 names
—
with ≥1 observation
Retail basket
—
with ≥1 observation
Days collected
—