/thesis-tracker

description: Write, save, and re-check an investment thesis — KPIs with thresholds, invalidation triggers, catalysts, a pre-mortem, and a decision log; `--update` re-reads the saved file against new data and returns INTACT / WEAKENED / BROKEN

Thesis Tracker — Write It Down, Then Check It

⚠️ Data Verification — Do This Before Any Analysis

Before running any analysis, always retrieve the latest market data for the ticker:

1. **Fetch current price** — use web search or ask the user for the live price, 52-week range, and market cap. Never assume a price from training data. 2. **Confirm key figures** — the current value of every KPI the thesis names (revenue growth, margin, FCF, net debt, subscriber count, …). A thesis check against stale KPIs is worse than no check: it produces false confidence. 3. **State your data source** — fill in the `Data & Sources` header (next section) so the origin, as-of date, retrieval path, and confidence of every figure are explicit at the top of the output. 4. **Flag stale data explicitly** — if live data is unavailable, display this warning before proceeding:

> ⚠️ **Live data unavailable.** The following analysis uses training-data estimates which may be significantly out of date. Verify all prices and metrics before making any decisions.

Never silently substitute training-data estimates for current prices. When in doubt, ask the user to paste the latest quote.

---

📋 Data & Sources Header — Open Every Output With It

The first thing in the output is this provenance block, filled in — never left as placeholders. It is the standard documented on the [Data & Accuracy](https://yennanliu.github.io/InvestSkill/data-and-accuracy.html) page and the first thing `result-validator` looks for:

Data & Sources
  As of:      <date the figures represent, e.g. 2026-06-30>
  Source:     <primary docs — SEC EDGAR 10-K/10-Q, company IR, FRED, exchange data …>
  Retrieval:  <pasted by user | web/tool retrieval | model memory>
  Confidence: <HIGH | MEDIUM | LOW>

• `Retrieval: model memory` must be paired with `Confidence: LOW` — memory is a placeholder until confirmed against a primary source.
• Mixed sources: list each with its own as-of date rather than blending them.
• Sources disagree on the same figure (a closing price, a deal term): prefer the primary source (SEC filing, company release, exchange data) over an aggregator, then the most recent dated one. Show both values with their sources and dates, say which you used and why, and never average them.
• Data the user pasted is reported as `pasted by user`; do not upgrade its confidence beyond what the user's own source supports.

---

Overview

Every other framework in the catalog answers a question *today*: is it cheap, is the moat real, is the chart constructive, what does the bear say. None of them remembers the answer. Three months later the investor is left with a position, a vague feeling, and no record of what they expected to happen — which is exactly the state in which people average down on a broken story or sell a working one on a bad week.

**This skill turns an analysis into a contract with your future self.** It produces a single saved file per ticker that states, in falsifiable terms, *why* the position exists, *which numbers* have to keep being true, *what* would prove the idea wrong, and *when* the next check is due. Then, on `--update`, it re-reads that file against current data and hands back one word — **INTACT**, **WEAKENED**, or **BROKEN** (or **INSUFFICIENT EVIDENCE** when the data to decide is not there) — together with the specific line that changed.

It fills the gap between Playbook steps 1 (write the thesis), 9 (monitor), and 10 (know when to sell) in the Learning track, which today have no skill behind them.

**What it is not.** It is not a valuation, a screen, or a timing tool. It imports its KPIs and triggers *from* the other skills (`stock-eval` for quality metrics, `stock-valuation` for the fair-value range, `bear-case` for the thesis-killers, `catalyst-calendar` for dates, `position-ladder` for the share-count ceiling) and holds them to account. Run the analysis skills first; run this one to write down what they concluded.

---

1. Modes

| Mode | Invocation | What it does | |------|------------|--------------| | **Open** | `thesis-tracker NVDA` (+ pasted analyses, or none) | Interview the user, draft the thesis file, save it to `output/thesis/NVDA.md` | | **Update** | `thesis-tracker NVDA --update` (+ current data or a fresh analysis) | Re-read the saved file, refresh every KPI, re-test every trigger, return INTACT / WEAKENED / BROKEN (or INSUFFICIENT EVIDENCE), append to the decision log | | **Review** | `thesis-tracker --review` (no ticker) | Read every file in `output/thesis/`, list status and next-check date per ticker, flag anything overdue or sitting at INSUFFICIENT EVIDENCE | | **Close** | `thesis-tracker NVDA --close "sold — trigger 2 fired"` | Mark the thesis closed with the exit reason and final P&L; keep the file for `trade-postmortem` |

If the environment cannot write files, output the complete thesis file as a fenced block and tell the user to save it at the stated path — the *contract* matters, not the storage.

---

2. Inputs

| Input | Required | Default if unstated | |-------|----------|---------------------| | Ticker | ✅ for Open, Update, Close · not used by Review (it reads every file) | — | | The thesis in the user's own words | ✅ for Open | ask — never invent a thesis for the user | | Entry price and date, position size | recommended | record as "not stated" | | Prior analyses (`stock-eval`, `bear-case`, `stock-valuation`, …) | optional | derive KPIs from the thesis text | | Horizon | optional | 12 months | | Existing thesis file | ✅ for Update / Close | — |

**The user writes the thesis; the skill sharpens it.** If the stated thesis is not falsifiable ("great company, will go up"), push back with the three questions in Phase 1