Data Ingestion & Scoring Methodology | Stock Insider AI | Stock Insider AI

1. SEC Ingestion & Normalization

Our automated system monitors the Securities and Exchange Commission (SEC) EDGAR RSS feeds and database indices 24/7. When a corporate insider files a **Form 4 (Statement of Changes in Beneficial Ownership)**, our pipeline pulls the raw XML filing within seconds.

Once retrieved, we normalize the raw data:

Ticker Normalization: Mapping the reported issuer CIK (Central Index Key) to active trading ticker symbols.
Title Normalization: Raw filing descriptions (e.g., "Chief Executive Officer & President", "CEO") are normalized into standardized roles (CEO, CFO, Officer, Director, 10% Owner).
Transaction Identification: Isolating open-market purchases (Code P) and sales (Code S) from routine employee option grants (Code A) or option conversions (Code M) which carry less sentiment weight.

2. Insider Confidence Score Calculation

The **Insider Confidence Score (1-100)** is our proprietary metric designed to gauge executive conviction. The score starts at a neutral baseline of 50 and is adjusted dynamically using the following weights:

+15

Strong Buy Ratio

Applied if purchases represent more than 70% of total insider transaction volume in the past 90 days. Conversely, a heavy sales ratio drops the score by up to 15 points.

+20

Cluster Buying Detection

Triggered when 3 or more unique corporate insiders purchase shares on the open market within a 30-day window. Multiple buyers indicate shared strategic alignment.

+15

CEO Buying Intensity

Calculated when the Chief Executive Officer purchases stock directly. CFO buying adds 10 points due to the CFO's direct visibility into treasury operations.

3. Data Integrity & Verification

To verify our calculations, every transaction is directly linked to its source filing on the SEC's official website (sec.gov). Users can audit the raw Form 4 filing PDF/XML page at any time by clicking the "Filing" external link on our transaction tables.