Quantitative Signals & Scoring

Sigma score

A composite 0 to 100 score that ranks an insider transaction by how predictive it is of forward returns, blending conviction, role seniority, cluster strength and timing into one number.

A single Form 4 or MAR Article 19 filing is noisy. The Sigma score collapses several weak, partially-correlated factors into one cross-sectional rank so that the top decile can be traded as a portfolio. Each factor is first turned into a z-score within its peer group, then weighted and squashed back into the 0 to 100 band.

Conviction: trade value relative to the insider's known net worth and the company's free float.
Role: a CEO or CFO buy outranks an independent director's, which outranks a 10% holder's.
Cluster: several insiders buying within a short window lifts the score sharply (see cluster signal).
Timing: proximity to earnings and recent price weakness are both scored.

Formula

Sigma = 100 \cdot Φ (i = 1 \sum n w_{i} z_{i})

w_i: weight of factor i (sums to 1)
z_i: peer-group z-score of factor i
Φ: standard normal CDF, mapping the blend into [0,1]

InsidersTradesSigma

Quantitative Signals & Scoring

Sigma score

A composite 0 to 100 score that ranks an insider transaction by how predictive it is of forward returns, blending conviction, role seniority, cluster strength and timing into one number.

Conviction: trade value relative to the insider's known net worth and the company's free float.
Role: a CEO or CFO buy outranks an independent director's, which outranks a 10% holder's.
Cluster: several insiders buying within a short window lifts the score sharply (see cluster signal).
Timing: proximity to earnings and recent price weakness are both scored.

Formula

Sigma = 100 \cdot Φ (i = 1 \sum n w_{i} z_{i})

w_i: weight of factor i (sums to 1)
z_i: peer-group z-score of factor i
Φ: standard normal CDF, mapping the blend into [0,1]

Related terms

Related terms