How it works

The methodology behind every study.

Every number in a Deepfolio study is sourced from a real SEC filing. Every verdict is the output of a documented framework, not a chatbot guess. This page explains the full pipeline end-to-end.

01 — Verdict framework

The four-verdict taxonomy.

Every Deepfolio study assigns one of four verdicts based on ROIC trajectory, FCF quality, competitive position durability, and capital allocation discipline — adapted from the frameworks of Damodaran, Mauboussin, and Greenblatt.

Compounder

ROIC sustainably above cost of capital, widening moat, disciplined capital allocation. The target holding type.

Cyclical

Earnings move with macro or commodity cycles. Not structurally impaired; requires cycle timing and balance-sheet tracking.

Turnaround

Returns currently below cost of capital but operational or structural change creates a credible path to recovery.

Decliner

Structural revenue or margin erosion with no clear recovery thesis. Capital at risk without a catalyst.

Verdicts are assigned at the section level (e.g. Competitive Position) and at the overall study level. They do not imply a price target or a buy/sell recommendation.

02 — Data pipeline

Four stages, zero AI-generated numbers.

Stage 1

Fetch

10-K, 10-Q, DEF 14A (proxy), and earnings call transcripts pulled directly from SEC EDGAR and official company sources.

Stage 2

Extract

Structured financials (income statement, balance sheet, cash flow) extracted from XBRL tags — not AI parsing. Numbers are exact as filed.

Stage 3

Reason

AI analysis runs 4 passes over the filing text: business model, economics, competitive position, and capital allocation. Each claim cites the source document.

Stage 4

Deliver

9-section study with verdict framework applied. KPIs calculated from raw financials — never from AI estimates or third-party aggregators.

03 — What we don't do

Scope boundaries — explicit.

Deepfolio is a fundamental research tool. The following are intentionally outside scope:

✕Price targets — we don't produce DCF outputs, fair value estimates, or buy/sell ratings.

✕Technical analysis — no chart patterns, momentum signals, or price-based indicators.

✕Options flow & sentiment — no short interest, put/call ratios, or social media signal.

✕Earnings estimates — all numbers are historical (as-filed), never forward projections.

✕AI-hallucinated figures — every KPI is computed from raw XBRL data. AI writes prose; it never writes numbers.

✕Investment advice — Deepfolio is educational research. Always verify against primary EDGAR filings before acting.

04 — Coverage

What’s supported and what isn’t.

Supported: US-listed equities (NYSE, NASDAQ, AMEX) with at least two annual 10-K filings on SEC EDGAR. Studies cover up to 15 years of financial history depending on availability.

Not yet supported: OTC / pink sheet securities, foreign private issuers (20-F filers), REITs with non-standard segment reporting, BDCs, SPACs, and shell companies with no operating history.

Currently 7 public sample studies are available below. Run any covered ticker on the homepage for a full 9-section study.

Ticker	Company	Sector	Data span	10-Ks	10-Qs
AAPL	Apple Inc.	Technology	5yr	3	6
GOOGL	Alphabet Inc.	Communication Services	5yr	3	6
MSFT	Microsoft Corporation	Technology	4yr	3	6
MU	Micron Technology, Inc.	Technology	5yr	3	6
NVDA	NVIDIA Corporation	Technology	5yr	3	6
RIG	Transocean Ltd.	Energy	5yr	3	6
TSLA	Tesla, Inc.	Consumer Cyclical	5yr	3	6

Try it

Run your first study — free.

Free tier includes 2 studies with no credit card required. Analyze any US-listed ticker.

Analyze a ticker →