α
AlphaSteve
← Brain index

Deliverable Suite — Standard Outputs Per Thesis

Each piece of analytical work produces a structured set of artifacts, not just a thesis note. The thesis is the analysis; the suite is the operational discipline that makes the analysis reviewable, shareable, and calibratable over time.

Important distinction. A thesis is research output, not a position. The thesis library (09-Theses/) is the workbench. Portfolio entries (12-Portfolio/Portfolio.md) and watchlist entries (11-Watchlist/Watchlist.md) are the operational containers downstream of the verdict:

  • Verdict = buy → row added to Portfolio, transaction logged, position managed live
  • Verdict = pass-with-trigger → row added to Watchlist, monitored against trigger price
  • Verdict = pass / avoid → thesis filed in workbench; no further action unless conditions change

Position management (sizing, sell triggers, kill-criteria monitoring, P&L) lives in the portfolio register, not the thesis. The thesis is the frozen reasoning; the portfolio is the live operations.

Tier 1 — Mandatory for every thesis

These three artifacts are produced for every name AlphaSteve analyzes — pass, hold, buy, or short. Skipping any of them is breaking the discipline.

1. Thesis note — 09-Theses/<TICKER>.md

The long-form analytical document following investment-thesis-template. Fully wiki-linked into the skill library. Source of truth for the reasoning.

Required sections: bottom line, business, structural assessment, variant perception, valuation triangulation, quality and management, macro/cycle, risk/pre-mortem, position sizing, what I don't know, next review.

Length: typically 6-10 pages for buy theses; 3-5 for passes.

2. Dashboard visualizer — 09-Theses/<TICKER>-dashboard.html

A single-page HTML dashboard distilling the thesis to its essential visuals: recommendation pill, four key numbers, valuation triangulation chart, reverse-DCF sensitivity grid, scenario probabilities, kill criteria, key signal callouts (insider activity, narrative-cycle phase).

Purpose:

  • 60-second re-read at each calibration checkpoint
  • Cover sheet when sharing the thesis
  • Anchor for variant perception in one image

Format: self-contained HTML — opens in any browser or in Obsidian with the HTML preview plugin. Inline styles, no external dependencies except CDN-hosted Chart.js.

3. Calibration tracker — 10-Calibration/<TICKER>.md

The discipline that turns a thesis into a learning loop. Records:

  • Initial prediction snapshot (price, central value, scenario probabilities)
  • Specific testable predictions with timelines
  • Kill criteria with checkbox states
  • Re-read schedule (quarterly + 12-month + 24-month)
  • Checkpoint log (appended at each scheduled re-read)
  • Calibration scorecard (built up over multiple checkpoints)

Template: see _TEMPLATE.md in the same folder.

Tier 2 — Build when warranted

These artifacts are produced when the analysis calls for them — not mandatory, not optional, but situational.

4. Valuation spreadsheet — 09-Theses/<TICKER>-model.xlsx

When to build:

  • Cyclical names where mid-cycle normalization requires modeling
  • Conglomerates / multi-segment businesses where SOTP needs careful build
  • Names with material capital structure complexity
  • Names with multiple sensitivity dimensions that benefit from interactive exploration
  • Any high-conviction position (Core 1 / Core 2 sizing)

Required sheets:

  • Assumptions input (drivers: revenue growth, margins, capex, working capital, capital structure, WACC, terminal)
  • P&L forecast (5-10 years, mid-cycle normalized)
  • EPV calculation
  • Reverse-DCF (current EV solved for required growth/margin)
  • Forward DCF (cross-check)
  • Comparables table
  • Sensitivity grids (revenue CAGR × terminal multiple; margin × growth; WACC × terminal growth)
  • Scenarios (bull/base/bear with probability weighting)

The spreadsheet is the live model that gets re-run at each calibration checkpoint as new quarters print.

5. Pitch deck — 09-Theses/<TICKER>-deck.pptx

When to build:

  • Theses being shared / presented to other capital
  • High-conviction longs (Core 1)
  • Names being pitched into a fund / committee process
  • Educational / training use

Required sections: situation overview, business, variant perception (most important slide), valuation triangulation, scenarios, kill criteria, what we don't know, recommendation.

Length: 12-20 slides typically. Two slides per thesis section, not more.

6. Primary-research log — 09-Theses/<TICKER>-research-log.md

When to build:

  • Any conviction long (Core 1 / Core 2)
  • Any name where Tier 1-2 evidence is doing material analytical work
  • Any name with ongoing positions (research updates over time)

Format: dated entries of:

  • Channel checks (vendor / customer / distributor conversations)
  • Expert calls (within compliance constraints)
  • Ex-employee or industry observer conversations
  • Customer product use observations
  • Regulatory filings beyond the standard 10-K/10-Q (court documents, FOIA results, patent filings, hiring patterns)
  • Conference notes

Each entry tier-ranked against the 04-intellectual-virtues evidence framework. Source ranking matters more than volume.

Note: a standalone skill for primary research methodology is on the kit's improvement list per kit-debrief-001-PLTR section 'Gap 2'.

Tier 3 — Optional / situational

7. Counter-thesis brief — 09-Theses/<TICKER>-counter.md

For high-conviction longs (Core 1), an explicit short / pass writeup written from scratch (not just the bear section of the main thesis). The discipline of building the bear case independently surfaces what the embedded version misses.

8. Peer-comp tearsheet — 09-Theses/<TICKER>-comps.md (or .xlsx)

For sectors where comparables analysis dominates (financials, REITs, asset managers, MLPs, mining). A separate document tracking the peer set, their multiples, and the target's spread vs. peers through time. Updated quarterly.

9. Event calendar — 10-Calibration/event-calendar.md

Cross-portfolio, not per-thesis. A tracked list of upcoming earnings dates, regulatory rulings, patent expiries, contract renewals, debt maturity walls for every active position. Aggregated calendar view of what's on the radar.

Output discipline

For every thesis, the filing pattern is per-thesis bundles under 09-Theses/<TICKER>/:

09-Theses/
└── <TICKER>/
    ├── <TICKER>.md                  (Tier 1: thesis)
    ├── <TICKER>-dashboard.html      (Tier 1: dashboard)
    ├── <TICKER>-calibration.md      (Tier 1: calibration tracker — lives with the thesis)
    ├── <TICKER>-model.xlsx          (Tier 2: when warranted)
    ├── <TICKER>-deck.pptx           (Tier 2: when warranted)
    ├── <TICKER>-research-log.md     (Tier 2: when warranted)
    ├── <TICKER>-counter.md          (Tier 3: optional)
    └── <TICKER>-comps.md            (Tier 3: optional)

10-Calibration/
├── _calibration-template.md         (copy into 09-Theses/<TICKER>/ for each new name)
├── kit-debrief-XXX-<TICKER>.md      (vault-level meta — debriefs about the kit itself)
└── event-calendar.md                (Tier 3: aggregated across portfolio)

The per-thesis bundle keeps everything for one name in one folder — the analyst working a thesis has the dashboard, model, and calibration tracker all together. The 10-Calibration folder holds vault-level meta: the template, kit-improvement debriefs, and any cross-portfolio aggregations.

Why the discipline matters

The point of the suite is not paperwork. Each artifact serves a specific purpose in the analytical workflow:

  • The thesis is the reasoning — the substantive output.
  • The dashboard is the re-read anchor — what you look at first at each checkpoint to decide whether deeper work is needed.
  • The calibration tracker is the learning loop — without it, you'd never know whether your reasoning is calibrated.
  • The spreadsheet is the live model — quarterly updates require re-running, not rebuilding.
  • The deck is the shareable artifact — clarifies the thinking by forcing it into a different format.
  • The research log is the evidence base — without it, "I remember someone said" replaces source ranking.
  • The counter-thesis is the cognitive defense — protects against confirmation bias.
  • The event calendar is the radar — prevents missing a known catalyst.

Each artifact reduces a specific failure mode. Together they make the workflow durable across time, market regimes, and analyst memory.

Linked