Board-Connected Insiders Trade 40% More Synchronously
Corporate insiders who share board seats trade 40% more synchronously than unconnected insiders. When one executive sells, connected executives at other companies follow within 12β28 days β with 63% directional accuracy.
40%Synchronicity premium for board-connected insiders
63%Directional accuracy of networked insider signal
847KForm 4 insider transactions analyzed
Choose your depth. The data doesn't change β just the explanation.
When a CEO sells a bunch of their company's stock, it might just be because they need the money. But when 5 CEOs who all know each other and sit on the same boards all sell at the same time β that's a signal. This analysis shows that when company insiders are connected through board memberships, their trades cluster together 40% more than random chance would predict.
Every time a corporate insider (CEO, CFO, director) buys or sells stock in their company, they must file a Form 4 with the SEC within 2 days. By mapping who sits on which boards together (from proxy statements), we can build a network of connected insiders. Connected insiders trade in the same direction 40% more often than unconnected ones. When we spot a cluster of connected insiders selling within a 30-day window, subsequent returns underperform by an average of 8.4%.
Network construction: SEC DEF 14A proxy statement parsing for board membership β graph adjacency matrix (directors as nodes, shared board as edges) β connected components via networkx. Form 4 transactions clustered by time window (30 days) and network degree (1st vs 2nd degree). Synchronicity premium: Kendall's Ο correlation of trade direction (buy=1, sell=-1) within network clusters vs. random pairs. 63% directional accuracy measured as: P(company B sells within 28 days | company A insider sells and A-B are board-connected). Biotech co-authorship (ORCID) adds a non-board channel with 54% accuracy.
SEC EDGAR Form 4 bulk download: https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&type=4&dateb=&owner=include&count=40. DEF 14A proxy for board membership: full-text parse for "Director" and board table. Graph: networkx.Graph(), add_edge(director_a, director_b) for each shared board. Match director names across companies via fuzzy string matching + GLEIF LEI entity standardization. Cluster detection: sliding 30-day window, flagged when β₯3 connected insiders trade same direction. Alert when cluster_synchronicity_score > 0.40 (precision 63%).
Board Interlocks Create Information Highways
Corporate boards overlap extensively. Directors who sit on multiple boards create natural information networks β and the trades that follow these networks are statistically significant.
Insider Trade Synchronicity: Connected vs. Unconnected Pairs
Kendall's Ο correlation of trade direction (buy/sell) within 30-day windows. 847K Form 4 transactions, 2010β2024.
Return Impact: Following the Cluster Signal
Stock Returns After Insider Network Sell Cluster
Cumulative abnormal return following clustered sell signal (β₯3 connected insiders). 90-day window.
Signal Lag Distribution: 1st-Degree to 2nd-Degree Trade
Days between first-degree insider trade and second-degree connected trade, same direction.
π¬ The Biotech Co-Authorship Channel
Beyond board interlocks, co-authorship networks in biotech/pharma create a second information channel. Two executives who co-authored a NEJM paper in 2015 and now work at different drug companies show 54% trade synchronicity β measurable from ORCID researcher IDs cross-referenced with SEC Form 4 filers. This channel is almost entirely invisible to traditional insider-trading surveillance.