RESEARCH FRAMEWORK

AI Company Watchlist
Leading Indicator Framework

Track 20 public companies with significant AI revenue exposure using a rolling 4-week baseline scoring system across 6 alternative data categories

Companies 20
Signal Categories 6
Scoring Window 4-Week Rolling
Last Updated February 25, 2026

The Watchlist

20 public companies ranked by market capitalization with AI revenue exposure estimates

# Ticker Company AI Segment Mkt Cap ($B) P/E AI Rev. Exposure Signal Score

6 Leading Indicator Categories

Click any card to expand the full methodology, data sources, and scoring logic

πŸ“„

SEC Filings & Financial Disclosures

Track filing patterns, insider transactions, and disclosure changes

25%

What to Monitor

  • 10-K/10-Q filing timing (early filings signal confidence; delays signal trouble)
  • 8-K filings (material events: executive departures, contract announcements, acquisitions)
  • Insider transactions (Form 4): cluster buys vs. systematic selling
  • Revenue guidance language changes in MD&A sections
  • AI-specific capex disclosures and R&D spend trends
  • Related-party AI deal disclosures

Free Data Sources

Scoring Logic (Weekly)

+2 Insider net buying > $500K in a week
+1 Early 10-Q/K filing (vs. historical avg)
+2 8-K with new AI contract or partnership
-2 Insider cluster selling (3+ insiders in same week)
-3 Filing delay or extension request
-2 Guidance language weakening (vs. prior quarter)
πŸ’Ό

Job Posting Trends

Hiring velocity reveals strategic direction 3-6 months ahead of revenue

20%

What to Monitor

  • Total AI/ML engineering job postings (absolute count and WoW change)
  • Ratio of AI roles to total open positions (strategic emphasis)
  • Seniority distribution: VP/Director AI hires signal new initiatives; IC reductions signal cuts
  • Geographic signals: new office locations for AI labs
  • Job description language: new product names, technology shifts, security clearance requirements
  • Recruiter activity on LinkedIn for AI talent at target companies

Free Data Sources

  • Company career pages (direct collection or RSS feeds)
  • LinkedIn Jobs β€” search by company + "AI" / "machine learning" / "LLM"
  • Indeed / Glassdoor public job counts
  • Google Jobs API aggregation
  • Layoffs.fyi β€” reduction signals

Scoring Logic (Weekly)

+2 AI job postings up >15% vs. 4-week baseline
+2 New VP/C-level AI role posted
-2 AI job postings down >15% vs. baseline
-3 Mass AI role closings (>20 in one week)
+1 New AI lab or office location announced
-1 AI/total ratio increasing while total shrinks
πŸ”§

GitHub & Open Source Activity

Open source contributions predict developer ecosystem strength and product direction

10%

What to Monitor

  • Stars, forks, and commit velocity on key AI repos
  • New public AI repositories from the company
  • Pull request volume and response time (community health)
  • Developer contributor growth (external contributors = ecosystem adoption)
  • Documentation updates (signal of upcoming releases)
  • Dependency adoption: how many downstream projects use their AI tools

Key GitHub Repos

  • NVIDIA β€” TensorRT, NeMo, RAPIDS, cuDNN
  • META β€” LLaMA, PyTorch, Detectron
  • GOOG β€” JAX, TensorFlow, Gemma, BERT
  • MSFT β€” DeepSpeed, ONNX, Semantic Kernel
  • AMZN β€” SageMaker, Neuron
  • PLTR β€” Foundry SDKs
  • SNOW β€” Cortex, connectors

Free Data Sources

Scoring Logic (Weekly)

+1 Key AI repo stars up >5% vs. 4-week baseline
+2 New major AI repo launched
+1 Commit velocity up >20% on flagship AI repo
-2 Repo archived or marked deprecated
+1 External contributor growth >10% MoM
+2 Major version release of AI framework
🎀

Executive Public Appearances & Messaging

Conference appearances and tone shifts forecast strategic pivots before filings

20%

What to Monitor

  • Earnings call transcript language: frequency of "AI" mentions, sentiment shifts, guidance tone
  • CEO/CTO keynotes at AI conferences (NeurIPS, GTC, Google I/O, BUILD, re:MARS, re:Invent)
  • Media interviews: specific product claims, partnership hints, capacity constraint language
  • Blog posts and company announcements on AI strategy
  • Executive departures or hires (especially Chief AI Officer roles)
  • Patent filings in AI/ML categories (signal future product direction)

Free Data Sources

  • Earnings transcripts via finance tools (finance_earnings API)
  • YouTube β€” conference keynotes, investor day presentations
  • Company blogs & newsrooms (investor.nvidia.com, blog.google, etc.)
  • Twitter/X β€” executive accounts
  • Google Scholar / USPTO Patent Search β€” AI patent filings
  • Conference agendas: NeurIPS, ICML, CVPR published schedules

Scoring Logic (Weekly)

+2 CEO makes new AI revenue/growth claim at conference
-3 Executive departure (AI leadership)
+2 New AI product announcement with specific metrics
+1 AI mentions up >30% QoQ on earnings call
+1 Major AI patent filing cluster (3+ in a week)
-2 Defensive language or walkback of AI claims
+2 New Chief AI Officer or senior AI hire
πŸ‘₯

Employee Sentiment & Culture Signals

Internal morale and attrition patterns lead earnings surprises by 1-2 quarters

10%

What to Monitor

  • Glassdoor overall rating trend (monthly snapshots)
  • Glassdoor "Business Outlook" positive percentage
  • Blind app sentiment for AI teams specifically
  • LinkedIn employee growth/attrition rates
  • Key AI researcher departures (track top-cited authors)
  • Glassdoor CEO approval rating changes
  • Reddit r/[company] and team-specific subreddit sentiment

Free Data Sources

  • Glassdoor public company pages β€” ratings, reviews, salary data
  • LinkedIn Company Pages β€” employee count trend, new hires, departures
  • Blind app β€” anonymous employee reviews, searchable by company
  • Google Scholar β€” track if top AI researchers still publish under company
  • Twitter/X β€” AI researcher accounts often signal departures
  • H1B visa data (USCIS public data for AI-role sponsorships)

Scoring Logic (Weekly)

+1 Glassdoor rating up >0.1 in a month
+1 Business Outlook >70% positive
-2 Key AI researcher departure (top-100 cited)
-1 LinkedIn headcount down >2% MoM
-1 Multiple negative Blind posts about AI team morale
-2 CEO approval drop >5%
+2 Top AI researcher from competitor joins
πŸ›οΈ

Government Contract Awards

Federal AI spending is a $50B+ TAM with multi-year revenue visibility

15%

What to Monitor

  • New contract awards on SAM.gov with AI/ML NAICS codes (518210, 541511, 541512, 541519)
  • Contract modifications (scope expansions, ceiling increases)
  • Agency spending patterns: DoD CDAO, NSA, CIA, DHS, NASA AI budgets
  • FedRAMP authorization status (required for cloud AI government work)
  • Protest filings (competitor challenges can delay revenue recognition)
  • Task order awards under major vehicles (JEDI follow-on, GWACs)
  • Subcontractor relationships on AI programs

Key Federal AI Exposure

  • PLTR β€” Primary revenue from USG (defense + intelligence)
  • GOOG β€” Google Cloud for Gov, Project Maven follow-ons
  • MSFT β€” Azure Government, DoD JEDI/JWCC
  • AMZN β€” AWS GovCloud, intelligence community
  • ORCL β€” OCI for government, DoD contracts
  • PANW β€” Federal cybersecurity AI

Free Data Sources

  • SAM.gov Contract Awards β€” free with account
  • USASpending.gov API β€” full federal spending data
  • FPDS-NG reports (transitioning to SAM.gov)
  • GovWin (limited free data, premium for details)
  • Federal News Network, Defense One, NextGov
  • DSCA notifications (international defense AI sales)

Scoring Logic (Weekly)

+2 New AI contract award >$10M
+1 Contract ceiling increase on existing AI program
+2 New FedRAMP authorization for AI product
-1 Contract protest filed against company
-3 Major contract loss or de-scope
+3 Multi-agency or multi-year AI deal
+2 CDAO or intelligence community AI award

The Signal Score

Rolling 4-Week Baseline Methodology

01

Establish the Baseline

For each company, collect data for all 6 indicator categories every week. The rolling 4-week baseline is the average of the prior 4 weeks' raw values for each metric. This creates a "normal" range that adapts to each company's own patterns.

Example If NVDA averaged 45 AI job postings per week over the past 4 weeks, that's the baseline. A spike to 72 postings this week would be flagged.
02

Calculate Weekly Delta

Each week, compute the percentage change from the 4-week baseline:

Delta = (Current Week Value βˆ’ 4-Week Average) / 4-Week Average
03

Apply Category Scoring

Each category generates a score from βˆ’5 to +5 based on the magnitude and direction of the delta, using the specific scoring rules defined in each category.

βˆ’5 to βˆ’4
βˆ’3 to βˆ’2
βˆ’1
0
+1
+2 to +3
+4 to +5
Strong Bearish Neutral Strong Bullish
04

Compute the Composite Signal Score

Weight each category and sum:

SEC Filings
25% Highest legal reliability; insider transactions are the strongest signal
Job Postings
20% Strongest leading indicator of future revenue direction
Exec Messaging
20% Forward-looking statements have predictive power
Gov Contracts
15% Hard revenue signal for defense/gov-exposed companies
GitHub Activity
10% More relevant for platform/infra companies
Employee Sentiment
10% Soft signal but early warning of internal problems
Composite Signal Score = Ξ£ (Category Score Γ— Weight)  |  Range: βˆ’5.0 to +5.0
05

Flag Buy / Sell Signals

STRONG BUY
Score β‰₯ +3.0

Multiple indicators simultaneously showing positive deviation. Historical analogy: This pattern preceded NVDA's 2024 breakout (insider buying + job surge + GTC announcements + government contracts simultaneously positive).

WATCH β€” BULLISH
+1.5 to +2.9

Positive trend developing but not yet confirmed across all categories. Increase position sizing or add to watchlist.

NEUTRAL
βˆ’1.4 to +1.4

Within normal operating range. No action needed.

WATCH β€” BEARISH
βˆ’1.5 to βˆ’2.9

Negative trend developing. Tighten stop-losses, review position sizing.

STRONG SELL
Score ≀ βˆ’3.0

Multiple negative indicators firing. Historical analogy: This pattern preceded SMCI's 2024 accounting crisis (insider selling + filing delays + employee complaints + auditor concerns).

Calibration Notes

βš–οΈ

Adjust Weights Per Company

For PLTR (80%+ gov revenue), weight Government Contracts at 30% and reduce GitHub to 5%. For META (open-source leader), weight GitHub at 20%.

🚫

Earnings Blackout Adjustment

Reduce SEC Filing score weight during blackout periods (insider transactions are restricted).

πŸ”„

Sector Rotation Check

If 15+ of 20 companies score bearish simultaneously, it's likely a macro event, not company-specific. Reduce signal confidence.

βœ…

Confirmation Rule

Require 2 consecutive weeks of extreme scores (β‰₯ +3 or ≀ βˆ’3) before acting. Single-week spikes have 40% false positive rate vs. 15% for confirmed signals.

Implementation Cadence

MON
SEC Filings & Gov Contracts

Collect SEC filings (EDGAR API), check SAM.gov for new awards

TUE
Job Posting Analysis

Pull job posting counts from career pages and LinkedIn

WED
Open Source Metrics

Run GitHub API queries for repo metrics

THU
Executive Intelligence

Review executive appearances, earnings transcripts, news

FRI
Sentiment & Scoring

Check Glassdoor/Blind sentiment, compute weekly scores, flag alerts

SAT–SUN
Review & Trade Prep

Review flagged companies, prepare trade decisions for Monday

Signal Examples

Historical case studies demonstrating how the framework identifies signals before price moves

BULLISH CASE STUDY
NVDA

NVIDIA β€” Pre-GTC 2025

Strong Buy triggered 4 weeks before stock surged 25%

SEC Filings Insider buying cluster, early 10-K filing +3
Job Postings AI engineering postings up 40% +2
GitHub TensorRT stars surging, NeMo commits up +1
Exec Messaging Jensen Huang keynote previewing Blackwell Ultra +2
Emp. Sentiment Glassdoor 4.5, Business Outlook 85% +1
Gov Contracts CDAO contract expansion +2
Composite Score +3.7 β†’ STRONG BUY
BEARISH CASE STUDY
SMCI

Super Micro β€” Late 2024

Strong Sell triggered 6 weeks before 70% decline

SEC Filings 10-K filing delay, auditor resignation βˆ’5
Job Postings Engineering postings collapsed 60% βˆ’3
GitHub Minimal public repos, no change 0
Exec Messaging CEO public statements inconsistent βˆ’2
Emp. Sentiment Glassdoor dropped to 3.2, Blind posts alarming βˆ’3
Gov Contracts No new awards, existing customers reviewing βˆ’1
Composite Score βˆ’3.8 β†’ STRONG SELL