Value investing · research desk

Read a stock like a business owner, not a ticker.

Live data from Interactive Brokers and Finviz, run through a Warren Buffett quality & value scorecard, then an AI analyst writes the honest verdict — bull case, bear case, and what would change its mind. Every run is archived below.

Browse the reports How it works
The workflow

From ticker to verdict

Each report is produced by the same six-step pipeline. The data and scoring are deterministic; the judgment is written by Claude, grounded strictly in the fetched numbers.

data sources Ticker IBKR Finviz Scorecard Narrative Report
1

Input a ticker

A single symbol — e.g. AAPL, MU, KO — is all that's needed.

2

Fetch the data

Two local command-line tools pull price history from Interactive Brokers and a fundamentals snapshot from Finviz.

3

Normalize & score

Raw figures are typed and run through a config-driven Buffett scorecard — 12 criteria, each pass / warn / fail, weighted into an overall verdict.

4

Sentiment read

A separate panel captures what the Street thinks: analyst target & recommendation, institutional/insider ownership, short interest.

5

AI analyst narrative

Claude writes the value-investor judgment — thesis, bull, bear, key risks, and "what would change my mind" — using only the fetched numbers.

6

Archive the report

Everything is assembled into a self-contained HTML report and published to the history below.

The data

Two sources, two jobs

Interactive Brokers supplies the price picture; Finviz supplies the fundamentals. Both are reached through small, read-only local tools — not hosted APIs.

📈

Interactive Brokers

Price & market data
How it's reached: a read-only Python CLI (ib_async) talks to IB Gateway running locally over the TWS API socket at 127.0.0.1:4002. Not a cloud API — nothing leaves the machine, and it can only read, never trade.
  • Historical OHLCV bars — 1 year of daily data, the basis for the price chart, 52-week range, 50/200-day moving averages, trend and 1-year return.
  • Quotes — last / bid / ask (delayed on the paper account).
  • Contract details — symbol resolution across exchanges.
  • Option-chain parameters — expiries and strikes.
Paper account → quotes are delayed and labeled as such; historical bars are unaffected.
🧮

Finviz

Fundamentals snapshot
How it's reached: a Python CLI (requests + BeautifulSoup) reads the public quote page finviz.com/quote.ashx with a realistic browser User-Agent and a 1-hour on-disk cache. No official API — single-user, low-volume, polite.
  • Valuation — P/E, forward P/E, P/B, P/S, PEG, P/FCF, EV/EBITDA.
  • Quality & returns — ROE, ROIC, ROA, gross / operating / net margins.
  • Balance sheet — debt/equity, current & quick ratios.
  • Growth — EPS & sales, historical and forward estimates.
  • Ownership & sentiment — institutional, insider, short float; analyst target & recommendation; dividend yield & payout.
A single current snapshot (~70 metrics) — rich, but limited multi-year history; trends come from the IBKR price series.
The lens

The Buffett scorecard

Twelve criteria split business quality from price. Each is scored pass / warn / fail against configurable thresholds and weighted into a 0–100 score and verdict band. The point isn't the number — it's separating "is this a wonderful business?" from "is it cheap?".

Return on equity≥ 15%
Return on invested capital≥ 12%
Gross margin (moat)≥ 40%
Net profit margin≥ 10%
Debt / equity< 0.5
Current ratio≥ 1.5
EPS growth, past & next 5Y≥ 10% / 8%
P/E discipline< 18
Price / book< 3
Price / free cash flow< 20
PEG< 1.2
● quality  ·  ● valuation  —   A high score on a cyclical business often just means peak-cycle earnings. The AI narrative is where that judgment gets applied.
The archive

Report history

Every analysis run, newest first. Click any card to open the full self-contained report.