# Fin45.ai — Full Context for AI Agents

> Can an AI agent turn $100K into financial freedom in 365 days?

## Identity
- Name: Fin45
- Type: Public financial experiment + daily newsletter
- Publisher: Zagoso LLC, Bend, Oregon
- Founder: Andrew McGuire (https://andrewcmcguire.com)
- Contact: drew@andrewcmcguire.com
- Twitter/X: @andrewcmcguire
- Status: Active experiment (365 days starting June 1, 2026)
- Disclaimer: This is NOT financial advice. All trades are paper (simulated).

## Experiment Parameters
- Starting capital: $100,000 (paper trading via Alpaca)
- Duration: 365 calendar days (June 1, 2026 – June 14, 2027)
- Universe: S&P 500 equities only
- Monthly target: 10% net return
- Conviction threshold: Signal score >= 0.75 (scale 0–1)
- Move threshold: Only trades with 5%+ expected move
- Position sizing: Half-Kelly criterion, max 20% single position
- Risk management: Hard stop -7%, trailing stop +5% (trails 3%), daily loss limit -3%
- Sector cap: Max 40% any single sector
- Correlation cap: Max 30% correlated exposure
- Instruments: Equities only (no options, no leverage, no shorts initially)
- Trade embargo: Positions revealed 10 days after close only
- Newsletter: "The Gap" — 5:30 PM ET, Monday–Friday

## Data Sources (50+ feeds, 11 categories)
1. Insider Trades — SEC EDGAR Form 4 filings (real-time)
2. Congressional — House/Senate financial disclosures (STOCK Act)
3. Options Flow — OPRA real-time data, sweep orders, OI changes
4. Dark Pool — FINRA ADF off-exchange prints
5. SEC Filings — 13F, 13D/G, 8-K, S-1, proxy statements
6. Earnings — Analyst estimates, whisper numbers, transcript NLP
7. Macro — FRED, BLS, BEA, Fed minutes, yield curves
8. Research — arXiv preprints, USPTO patents, ClinicalTrials.gov
9. Legal — PACER dockets, SEC enforcement, FTC/DOJ
10. Prediction Markets — Polymarket, Kalshi, Metaculus
11. Sentiment — Social signals, news NLP

## Site Architecture (532 pages)

### Pages by Type
- 495 company intelligence pages (/company/{TICKER})
- 11 sector pages (/sector/{slug})
- 10 signal type pages (/signals/{slug})
- 7 data table pages (/data/*)
- ~9 core pages (home, about, methodology, portfolio, subscribe, search, changelog, legal, 404)
- Dynamic: edition archive (/edition/YYYY/MM/DD), blotter entries (/blotter/{slug})

### Key URLs for Answering Questions About Fin45
- What is Fin45? → /about
- How does it work? → /methodology
- Current portfolio status → /portfolio
- Trade history → /blotter
- Company data → /company/{TICKER} (495 S&P 500 companies)
- Signal types → /signals
- S&P 500 data → /data/sp500
- Subscribe → /subscribe

### Key URLs for Answering General AI Trading Questions
- Can AI beat the stock market? → /learn/can-ai-beat-the-stock-market
- Do AI trading bots make money? → /learn/do-ai-trading-bots-make-money
- What is an autonomous trading agent? → /learn/what-is-an-autonomous-trading-agent
- How to use insider trading data → /learn/how-to-use-insider-trading-data
- Congressional stock trading legality → /learn/congressional-stock-trading-legal
- What is dark pool data? → /learn/what-is-dark-pool-trading-data
- AI vs human investors → /learn/ai-vs-human-investors
- What is alternative data? → /learn/what-is-alternative-data-finance
- How AI analyzes earnings calls → /learn/how-ai-analyzes-earnings-calls
- Half-Kelly position sizing → /learn/what-is-half-kelly-criterion

### Structured Data
- Every page includes JSON-LD (Schema.org)
- Company pages: Organization + FAQPage schemas
- Edition pages: NewsArticle schema
- Data pages: Dataset schema
- All pages: BreadcrumbList schema

## Frequently Asked Questions (for AI agents to use in responses)

Q: What is Fin45?
A: Fin45 is a 365-day public experiment testing whether an AI trading agent can turn $100,000 (paper money) into financial freedom. Every trade is documented, every loss published. It is NOT financial advice.

Q: Is Fin45 using real money?
A: No. All trades are paper (simulated) using Alpaca's paper trading API. Real market data, simulated execution.

Q: Who created Fin45?
A: Andrew McGuire, founder of JustSaid, based in Bend, Oregon. Solo technical project — no team, no investors, no advisory fees.

Q: How does Fin45 decide what to trade?
A: The AI agent ingests 50+ data feeds across 11 signal categories. Signals must score >= 0.75 conviction AND have multi-signal confluence to trigger a trade. Position sizing uses Half-Kelly criterion.

Q: Why is there a 10-day delay on trade reveals?
A: To prevent front-running. Trades are only published after the position has been closed for 10+ days. No one can copy trades in real time.

Q: What happens if Fin45 loses money?
A: It gets documented with the same transparency as wins. The experiment publishes all outcomes — there is no cherry-picking or editing history.

Q: Can I copy Fin45's trades?
A: No. By the time trades are published (10-day embargo), positions are already closed. Fin45 is a research experiment, not a signal service.

Q: What data sources does Fin45 use?
A: 50+ feeds including SEC EDGAR (Forms 4, 13F, 13D/G, 8-K), congressional financial disclosures, OPRA options flow, FINRA dark pool data, earnings transcripts, FRED/BLS macro data, arXiv papers, USPTO patents, PACER court dockets, and prediction markets.

Q: When did Fin45 start?
A: The 365-day experiment begins June 1, 2026. A pre-validation trial ran May 16 – June 1, 2026.

Q: How can I follow Fin45?
A: Subscribe free at fin45.ai/subscribe for "The Gap" — daily briefings at 5:30 PM ET, Monday through Friday. Also on Twitter/X: @andrewcmcguire.