AI Investing Learning System

A personal system for learning AI/tech investing with stock-focused daily briefings that search across social media and news for what people are saying about your 100+ watchlist stocks.

What This Does

1. Stock-Focused Intelligence - Scans Twitter, Substacks, and Google News for mentions of your watchlist stocks
2. Daily Briefings at 5 AM EST - Get a Telegram notification with price movers and social sentiment
3. Just Reply - Respond naturally and your picks are auto-extracted and logged
4. Track Performance - See how your picks do over time
5. Learn by Doing - Dynamic educational questions based on the day's active stocks

Quick Start (5 minutes)

Option 1: Telegram Bot (Recommended)

1. Create your bot:

   • Open Telegram, search for @BotFather
   • Send /newbot, follow prompts, get your token

2. Configure:

   cp .env.example .env
   # Edit .env and add your TELEGRAM_BOT_TOKEN

3. Get your Chat ID:

   pip install -r requirements.txt
   python -m src.delivery.telegram_bot --get-chat-id
   # Send /start to your bot, it will reply with your Chat ID
   # Add TELEGRAM_CHAT_ID to .env

4. Test it:

   python -m src.delivery.telegram_bot --send-report

5. Enable daily automation:

   • Go to repo Settings → Secrets → Actions
   • Add TELEGRAM_BOT_TOKEN and TELEGRAM_CHAT_ID
   • Reports will be sent every weekday at 7 AM EST

Option 2: Email

See .env.example for Gmail, Resend, or SendGrid configuration.

Daily Workflow

Morning (5 AM EST)

• Get a Telegram notification with your daily briefing
• See which stocks are being discussed across Twitter, Substacks, and news
• Price movers highlighted with 1-day and 5-day performance
• Grouped by investment theme (AI Infrastructure, Nuclear, Robotics, etc.)

What You'll See

• Quick Movers - Stocks moving >2% with discussion activity
• Stock-by-Stock Breakdown - For each active stock:
  • Current price and performance
  • Your investment thesis reminder
  • Links to what people are saying (with sources)
• Learning Questions - Dynamic questions based on the day's activity

Respond Naturally

Just reply to the bot with your thoughts:

"Bullish on NVDA here, datacenter numbers were insane. The AI training demand is real. Also watching AMD to see if MI300 gets traction."

The bot automatically:

• Logs your full response
• Extracts stock picks: BUY NVDA, WATCH AMD
• Tracks them in your portfolio

Track Your Progress

Send /picks to see your active picks Send /performance to see your stats

Stock Coverage

Complete SOXX ETF Holdings (35 stocks)

All semiconductor stocks including:

• Top 10: AVGO, AMD, NVDA, MU, INTC, AMAT, QCOM, LRCX, MRVL, ASML
• Mid-tier: ADI, TXN, KLAC, NXPI, ON, MPWR, MCHP, ARM, GFS, SWKS
• Smaller: ENTG, TER, MKSI, QRVO, WOLF, LSCC, ACLS, AMKR, CRUS, SLAB, ALGM, RMBS, COHR, MTSI, DIOD

Plus AI/Tech Leaders

• Mega Cap: MSFT, GOOGL, AMZN, META, AAPL
• AI Software: CRM, NOW, PLTR, SNOW, MDB, DDOG, PANW, CRWD, ZS, NET
• AI Infrastructure: SMCI, VRT, DELL, HPE, ANET, CSCO, VST, CEG, EQIX, DLR
• AI Pure Plays: PATH, AI, SOUN, UPST, S

See config/sources.yaml for complete list with investment theses.

Content Sources

Substacks

Newsletter	Focus
SemiAnalysis	Semiconductor deep dives, AI infrastructure
Stratechery	Tech strategy, platform dynamics
Not Boring	Tech trends, company deep dives
Fabricated Knowledge	Semiconductor analysis

Podcasts

Podcast	Why Listen
All-In	Real-time takes from tech investors
Acquired	How great tech companies were built
Invest Like the Best	Professional investor frameworks
BG2 Pod	Bill Gurley + Brad Gerstner

Twitter/X

10 curated accounts for AI investing insights. See config/sources.yaml.

Bot Commands

Command	What it does
/start	Get your Chat ID for setup
/report	Get today's report on-demand
/picks	View your active picks
/performance	See your stats
(any message)	Logged + picks extracted

Learning Frameworks

The system teaches you to think about:

Valuation

• P/E vs growth rate (PEG ratio)
• EV/Revenue for growth companies
• Free cash flow yield
• Rule of 40 for SaaS

AI-Specific Questions

• What is the company's right to win in AI?
• Is this pick-and-shovel or direct AI bet?
• How defensible is the AI advantage?
• What's priced in at current valuation?

File Structure

Learning-ai-investing/
├── config/
│   └── sources.yaml         # All sources and 100+ stocks
├── data/
│   ├── picks.json           # Your tracked picks (auto-synced)
│   ├── question_responses.json
│   └── reports/             # Daily report archive
├── src/
│   ├── delivery/
│   │   ├── telegram_bot.py  # Telegram delivery + response parsing
│   │   └── email_sender.py  # Email delivery option
│   ├── fetchers/
│   │   ├── stock_search.py  # Social search across sources
│   │   ├── substack.py      # Substack RSS fetcher
│   │   ├── twitter.py       # Twitter/Nitter fetcher
│   │   └── podcasts.py      # Podcast feed fetcher
│   └── report_generator.py  # Stock-focused report builder
├── .env.example             # Configuration template
└── requirements.txt

Automation

Reports are automatically generated and sent via GitHub Actions:

• Every weekday at 5 AM EST
• Scans 100+ stocks for social mentions and news
• Sent to Telegram (or email as fallback)
• Archived in data/reports/

To enable:

1. Add secrets to your repo (Settings → Secrets → Actions)
2. Enable GitHub Actions (Settings → Actions → General)

Philosophy

This system is designed for active learning:

1. Read with purpose - Reports highlight what matters
2. Form opinions - Reply with your thoughts
3. Track outcomes - See if you were right
4. Iterate - Learn from wins and losses

The goal isn't to be right every time. It's to improve your thinking over time.

---

Built for learning AI/tech investing the hard way - by doing it.