Equity Valuation Reports

Public-information equity research with self-contained HTML memos and live Excel models. Both packages are designed for hedge-fund PM review: they separate calculation integrity from trade actionability and recommend no risk until the evidence and implementation gates clear.

Live site: jainabhishek.github.io/equity-valuation-reports

Current memos

	Alphabet (GOOGL)	Nvidia (NVDA)
Stance	WAIT FOR PROOF / NO POSITION	WATCHLIST / NO POSITION
Market data	$354.24, August 7 regular-hours last trade	$223.90
Primary scenario range	$273.91 / $334.68 / $398.41	$57.22 / $165.40 / $377.68
Decision reference	$335.42 (−5.3%) · illustrative expected value	$165.40 (−26.1%) · base DCF, no probabilities
Position size	0.0% · implementation gates open	0.0% · seven gates open
Alphabet DCF cross-check	$143.64 · 90% terminal value, not the catalyst	—
Memo	Open →	Open →
Model	Download	Download

Alphabet: selected thesis and decision rule

Alphabet's AI capex may convert into GAAP D&A faster than consensus EPS incorporates it. The thesis works only if EBITDA fails to absorb the charge, reported results force FY2027–28 EPS cuts, and the multiple responds. D&A alone is not the signal, and it is not the direct source of DCF downside because it is added back in FCFF.

The current result is WAIT FOR PROOF / 0.0% because:

• no broker-level quarterly D&A consensus or revision history has been frozen;
• the $15.01 FMP FY2027 EPS snapshot conflicts with a separately cited range;
• no observed print has yet produced the required EPS revision;
• borrow/carry, crowding/squeeze, executable options and hedge inputs are missing;
• the DCF remains dominated by terminal value and is used as a risk frame only.

The conditional initiation rule is explicit: actual D&A must reach or exceed the frozen path, EBITDA must offset less than half of the surprise, FY2027 consensus EPS must fall at least 5% within ten trading days, net downside must remain at least 20%, and every implementation gate must clear.

Alphabet model architecture

The workbook is a second, formula-driven implementation of the Python model. Its first visible tab is Cover, followed by a source ledger and the calculation stack:

1. Sources separates filed facts, market observations and analyst assumptions.
2. Drivers splits FY2026 into reported H1 and forecast H2.
3. WACC builds CAPM from the August 7 Treasury rate, beta/ERP assumptions, after-tax debt cost and market-value capital weights.
4. Share Bridge uses July 15 point-in-time common shares, ordinary dilution and full minimum mandatory-convertible dilution. Preferred is not deducted again.
5. Depreciation models quarterly capex vintages, asset mix, useful lives and commissioning lags. Opening-vintage and forecast-vintage D&A are separate.
6. Valuation excludes reported H1, discounts only the August 8 onward stub, and normalizes terminal reinvestment using g / terminal ROIC.
7. Variant carries the consensus conflict and a quarterly catalyst template.
8. Scenarios, Sensitivities, Decision and Checks are formula-linked; no utility-based sizing or hardcoded position recommendation remains.

The DCF uses explicit haircuts for restricted marketable equity and non-marketable investments instead of silently treating them as cash. FY2026 capex is $200bn, the midpoint of management's $195–205bn guide; FY2027 is a visible $230bn analyst assumption, +15%, consistent with management's direction that spending would increase significantly.

Nvidia: selected thesis and decision rule

The public-information work does not establish an investable Nvidia variant at the August 7 price. The base revenue path is inside the frozen FY2027 aggregator range. The stock/model disagreement is a duration argument: spot implies an 8.62% WACC on the base operating path versus the explicit 10.70% CAPM case, or a blunt 36% uniform uplift to base revenue. Neither is a differentiated near-term earnings call.

The current result is WATCHLIST / NO POSITION / 0.0% because:

• no broker-level consensus cohort or revision history has been frozen;
• customer-financing exposure within $18.6bn of Q1 private investments and $27bn of investment commitments is not quantified;
• channel, lead-time, backlog and cancellation evidence is absent;
• Q2 results are scheduled for August 26 at 2:00 p.m. Pacific; actuals and forward guidance are not filed at this cut;
• live options, short interest, borrow, crowding and portfolio factor data are missing; and
• the bear/base/bull DCF states span $57.23 to $377.69.

The conditional rule is symmetric. A long requires observable upside to a frozen consensus, cleared demand-quality work and at least 20% net underwritten return. A short requires a reported break in the base path, estimate revisions, controlled uncapped upside and at least 20% net downside after implementation.

Nvidia model architecture

The workbook is authored separately with @oai/artifact-tool and recalculated against the Python model. Its 14-sheet stack opens with Cover, Decision and Sources, then carries the calculation and support layers:

1. Drivers distinguishes filed facts, market observations and analyst inputs.
2. Revenue anchors FY2027 on Q1 actual revenue and the Q2 guide, and uses Nvidia's current Data Center / Edge Computing framework.
3. WACC builds CAPM from the 4.65% Treasury, 1.35x beta, 4.50% ERP, after-tax debt cost and market-value weights.
4. Depreciation separates opening D&A from formula-driven forecast capex vintages with short- and long-lived asset buckets.
5. Equity Bridge credits cash at par but haircuts public, private and equity-method stakes by scenario. Operating leases are disclosed without a second deduction from cash flows that already include lease expense.
6. Valuation subtracts reported Q1 CFO less capex from FY2027 FCFF and normalizes terminal reinvestment with g / terminal ROIC.
7. Scenarios treats bear, base and bull as unweighted states. No unsupported expected value or Kelly sizing remains.
8. Reverse DCF, Sensitivities, Decision and Checks expose what spot must assume, the duration risk, the seven open capital gates and a forced zero position.

Alphabet sources

Need	Source / treatment
H1 financials, D&A components, balance sheet, shares and preferred mechanics	Alphabet Q2 2026 Form 10-Q and filed earnings release
Capex guidance and technical-infrastructure mix	Alphabet Q2 2026 earnings call
Asset lives	Alphabet FY2025 Form 10-K
Risk-free rate	U.S. Treasury daily yield curve, 4.65% on August 7
Price and 30-day share ADV	Read-only Robinhood quote and fundamentals snapshots
FY2027 EPS / EBIT / EBITDA	Frozen FMP aggregator snapshot, explicitly flagged as a source conflict
Beta, ERP, terminal ROIC, asset split/lags, probabilities and multiples	Analyst assumptions, labeled at point of use

The June capital structure is kept precise: 86m common shares and $19.25bn of mandatory-convertible preferred were issued in June; the $40bn ATM had no sales through June 30; the August 6 debt filing was preliminary with amounts still blank as of August 8. Registered but unsold capacity is not described as authorized capital or inserted into the share count.

Nvidia sources

Need	Source / treatment
Q1 financials, investments, commitments, concentration, D&A and share count	Nvidia Q1 FY2027 Form 10-Q
Current reporting framework, Data Center detail and Q2 guide	Nvidia Q1 FY2027 filed earnings release
FY2026 historical revenue	Nvidia FY2026 Form 10-K
Q2 FY2027 event timing	Nvidia investor-relations event page; August 26 at 2:00 p.m. Pacific
Risk-free rate	U.S. Treasury daily yield curve, 4.65% on August 7
Price	Read-only Robinhood regular-hours snapshot; refresh before risk
FY2027–28 revenue consensus	Frozen FMP aggregator snapshot; broker provenance and revisions remain open
Beta, ERP, terminal ROIC, asset lives, haircuts and hurdle	Analyst assumptions, labeled at point of use

Build and verification

From build/:

python3 alphabet_pm.py              # Alphabet JSON, model and memo
python3 nvidia_pm.py                # Nvidia JSON and memo
node nvidia_workbook.mjs            # Nvidia formula workbook via artifact-tool
python3 render.py                   # Both PM memos
./.venv/bin/python workbook.py      # Both workbooks
./.venv/bin/python previews.py      # landing/social images
./.venv/bin/python verify.py        # independent formula recalculation

verify.py recalculates both Excel files with a formula engine and compares them with separate Python implementations. It checks dated FCFF periods, D&A, capital bridges, WACC, terminal mechanics, scenario values, sensitivity centers, zero-size gates, sheet architecture and prohibited legacy claims. The Nvidia workbook is also imported, traced, formula-error scanned and rendered one sheet at a time with @oai/artifact-tool before circulation.

Known limitations

• No channel checks, expert calls or alternative data.
• No broker-level quarterly D&A consensus, estimate revision history, short locate/carry, crowding, live option chain or hedge basis.
• Alphabet's capex asset split, placement lags and opening-vintage runoff are estimates; the opening D&A run-rate is held flat because remaining lives by historical vintage are not disclosed.
• The Alphabet DCF is highly terminal-sensitive and should not be treated as a precise price target.
• Nvidia's FY2027 stub uses reported Q1 CFO less capex as an FCFF proxy; replace it with the filed Q2 bridge after results.
• Nvidia's revenue states are consolidated analyst scenarios. The current Data Center / Edge framework does not yet have enough reported history for a bottom-up forecast.
• Nvidia's DCF is 48%–76% terminal value across states and is a duration risk frame, not an executable price target.

Analytical research on public information. Not investment advice, a recommendation, or a solicitation.