$ prism-api --info
Behavioral Signal
Infrastructure
Quarterly behavioral signals covering US Russell 3000+ and Japan TSE, derived from 57 psychometric dimensions scored on regulatory filings. Four signals. One raw panel. Delivered via CSV, API, or S3.
$ prism-api --list-products
Signal Suite
Prism operationalizes "tone from the top" into a quantitative, cross-sectional measure. All inputs are public documents. No web scraping, no PII, no MNPI.
Conduct Risk Index R3K
+21.72% alpha · Sharpe 1.50
US long/short alpha signal. Walk-forward Carhart validated on Russell 3000. t=5.66, 11/12 formation years positive, defensive beta.
Governance Risk Index TSE
+5.63% alpha · Sharpe 1.46
Japan long/short alpha signal. 4,398 TSE-listed firms scored on EDINET filings. JP Fama-French 4 adjusted, t=4.06 Newey-West.
Enforcement Risk Index R3K
AUC 0.839 · 13× separation
US regulatory enforcement probability. Predicts SEC and DOJ actions across 8,342 firms. Compliance screening and D&O insurance pricing.
Prosecution Risk Score S&P 500
AUC 0.944 · nested LOO CV
DOJ criminal prosecution probability. Temporal behavioral drift signal — firms losing prosocial tone in public communications.
Tier 2 — Raw Behavioral Panel
Full 57-dimension quarterly panel. 17,500+ companies, 2001–2026. Build your own signals.
$ prism-api --query enforcement-risk
Enforcement & Prosecution Risk
Firms whose public communication loses its prosocial and moral tone over time are the enforcement-prone ones. The signal is drift, not level.
PROSECUTION_AUC . 0.94Nested cross-validation on S&P 500. 495 firms, 24 prosecuted. Honest out-of-fold scores. nested CV, S&P 500
LEAD_TIME ...... ~2 quartersSignal detectable before DOJ enforcement action. before enforcement
SIGNAL_TYPE .... Behavioral driftFirms losing prosocial tone over time, not cross-sectional level. temporal, not cross-sectional
ORTHOGONALITY .. 94.7%Signal variance NOT explained by sector + log-assets. Strongly idiosyncratic. unexplained by sector + size
FEATURES ....... 15Comparable AUC with 29 total features vs 242 in full model (1/8 the feature count). parsimonious composite
BOOTSTRAP_CI .. [0.91, 0.97] 95% CI, N=1000
$ prism-api --methodology
How It Works
SIGNAL Behavioral scores from regulatory filings and earnings transcripts
METHOD Proprietary scoring methodology
UNIVERSE US Russell 3000+ and Japan TSE (17,500+ companies)
REBALANCE Annual
DIMENSIONS 57 behavioral constructs per company
SOURCES_US SEC EDGAR (10-K, 10-Q, DEF 14A) + earnings transcripts
SOURCES_JP TDnet/EDINET regulatory disclosures
$ prism-api --pricing
Access
Evaluation
Sample panel, case study, coverage map, and full validation deck.
Tier 1 — Alpha Signals
Conduct Risk Index, Governance Risk Index (TSE), Enforcement Risk Index, Prosecution Risk Score. Quarterly, via API, CSV, or S3.
Tier 2 — Raw Panel
Full 57-dimension quarterly panel. 17,500+ companies, US + Japan. Build your own signals.
Listed on Neudata · Eagle Alpha
Request Evaluation
Sample panel, case study, coverage map, and full validation deck.
prism@stronglight.io