Tap a row to edit it here. Stored in localStorage only — never in source code.
Behavioral Rules Status
AI Analysis
Phase 1
Phase 2
Verdict
Profile
Theme ▸
Growth ▸
Tracker ▸
Dependency ▸
Stock Setup
Live Market Data (Google Sheets bridge)
Document Upload
Each document type has its own extraction prompt and output caps. Drop each file in its matching slot — this is what keeps concall/quarterly reads from running away.
📄
Annual Report
Upload PDF • full audited financials
📊
Investor Presentation
Upload PDF • forward guidance, capacity, order book
📈
Quarterly Results
Upload PDF • ~5 headline metrics only
🎙️
Concall Transcript
Upload PDF • pure Q&A, ~5–10 forward numbers
📑
Other Documents
Screener.in export, CRISIL/ICRA rating, industry report, fund house note
Manual Override (Optional)
Override:
Gemini 2.5 Flash • Temperature 0.0 • JSON output
Part 2 — External Analysis (alternate path)
Skips the upload pipeline entirely. Copy the prompt, open ChatGPT / Claude / Gemini, attach all your documents (annual reports, concalls, presentations across quarters), run it, then paste the JSON it returns here. The external chat does the heavy multi-document reading — this tool still computes every deterministic gate, the verdict and the allocation locally.
Clean JSON → used directly (0 Gemini calls). Messy/narrative paste → 1 text-only Gemini call normalizes it.
Analysis Progress
Validating API key...
Preparing documents...
Phase A — Extracting raw facts from each document...
Phase B — Consolidating all documents into one verdict...
Calculating Phase 3 — Verdict & Allocation...
Diagnostic Log
AI Extraction — Source Pages
✓ AI analysis complete — scores auto-populated in Phase 1, 2 & 3. Review and override if needed.
Display-only view of the last analysis. Sections marked
🔒 TRACK 2 need extraction fields the pipeline
does not yet emit — adding them is a coupled change (prompt + Phase B read + data contract together).
Theme Cycle Layer — Step 0
Score the 6 vital signs for the theme this stock belongs to (not the company).
Stage auto-computes → Stage Multiplier scales final allocation.
Stage 2 (Acceleration) = ×1.0 (full capital). Stage 1/3 = ×0.5. Stage 4/5 = ×0 (no new entry / exit).
Your edge: catch themes early, allocate at Stage 1–2, harvest before the crowd wakes up.
Growth Engine + Valuation
Project EPS forward 3 years under Bull / Base / Bear growth + exit-PE assumptions.
Get forward PE (FY+1/+2/+3), PEG, target prices and the multibagger matrix.
Freeze the baseline after a Base Run — the Projection Tracker (next build) will measure every
future quarter against it. AI never predicts here: your assumptions, deterministic math.
Projection Tracker — Thesis Accountability
Closes the loop: what we expected (frozen trajectory) vs what management promised
(Commitment Ledger, auto-harvested from concall/presentation runs) vs what actually happened
(quarterly actuals). Divergence shading: green = tracking ·
amber = drifting · red = >15% below base.
Flags fire deterministically — the 3-miss rule, thesis-drift, add-window, profit rotation.
Dependency Map — this stock's exposure
Where the revenue comes from, and which shared forces (GoI capex · RBI · China ·
crude/coal · USDINR · monsoon · USFDA · bond yields · private capex · export demand) move the thesis —
and with which sign. Wire this stock to its nodes below; the Exposure page then shows
the portfolio-wide concentration. Triggers still harvest from the documents' own risk language.
A severity-3 headwind surfaces on the Verdict next to R3.
Analysis Library
Every analysed stock, persisted from the registry. Reopen replays the saved dossier — zero API calls, zero cost.
HC Bucket Weights
Bucket
Fund%
Flow%
Tech%
Risk%
Max
HC1
40%
20%
15%
25%
₹40K
HC2
30%
25%
30%
15%
₹25K
HC3
25%
20%
35%
20%
₹10K
HC4
20%
25%
35%
20%
₹8K
Portfolio Dependency Map
Your real risk isn't per-stock — it's the shared forces your whole book hangs off.
Set each node's current state once; it cascades to every wired holding with the
correct sign (a rate cut is a tailwind for an NBFC, a headwind for a bank's NIM). Sensitivity is
auto-seeded from each stock's sector and tightened by the metrics already extracted — edit any stock's
wiring from the Dependency ▸ tab in the Analyzer.
wires every registry holding to its sector's nodes + auto-sets sensitivity from extracted metrics
Google AI Studio API Key
Get your free key at aistudio.google.com → Get API Key → Create API Key Free tier: 15 req/min, 1,500 req/day — enough for personal use.
Cache & Diagnostics (v3.0.6)
After Phase A (PDF extraction) completes, observations are cached. If Phase B fails, the next run skips Phase A and jumps straight to Phase B — saving 90+ seconds per retry.
The diagnostic log is available during analysis (see the "Diagnostic Log" panel). Copy it on error instead of screenshotting.
Build Info
Version: v3.6.2
AI Model: Gemini 2.5 Flash
Temperature: 0.0
Output: application/json
Architecture: Single HTML · No server · BYOK
Storage: localStorage only
Privacy: PDFs uploaded to Google via File API for analysis, then removed
Portfolio Numbers (stored locally — never in code)
These numbers are stored only in your browser's localStorage. They are never saved in the source code file and are not visible in your GitHub repo.
Investor Profile
Broker: Zerodha
Goal: —
Proven XIRR: 30.5% (7 years)
Style: Thematic Early Identifier + Patient Accumulator