Why discount earnings at cost of equity instead of WACC?

EPS-based models flow value to the equity holder by construction; the relevant discount rate is the rate of return equity holders require, not the blended cost of capital. Discounting EPS at WACC double-counts debt benefit and produces overstated fair values, particularly for leveraged companies.

Why is the report bear-case-first?

Bear-case-first is a forcing function against confirmation bias and against the well-documented analyst tendency to anchor on the bullish narrative. Risks land in section 2, before any bullish analysis; valuation models are built before the bear case is written so the scoring is not anchored to either extreme.

How is the margin of safety chosen?

Dynamically, by business quality. Quality 8-10/10 carries a 15% margin of safety; 6-7/10 carries 20%; 4-5/10 carries 25%; below 4 carries 30% or more. The margin scales with the cost of being wrong, not as a constant.

What triggers a hard fail in the decision overlay?

An accounting-quality gate failure, elevated survivability risk, or a fair-value midpoint below current price with confidence at or above 65 — any one of those overrides the composite score and blocks a buy rating regardless of multiple compression.

← Methodology

§ Per-stock methodology v2.4

How every report is produced.

Every monthly report on StockMarketAgent.AI is produced by the same disciplined, 9-phase workflow. We classify the business archetype, calibrate the discount rate to that archetype, run accounting-quality gates before trusting earnings, build the required valuation models, stress-test the result, and only then write the narrative — risks first. The goal is to separate the quality of the business from the attractiveness of the price, and to make every load-bearing assumption auditable.

§ One paragraph · 18 languages

Same numbers. Same rating. Eighteen voices.

Below is one real Section 2 risk paragraph from a published MSFT report, rendered in four languages at once. The figures, timeline, and rating glyphs are byte-identical across panels — only the prose is translated. Reports ship in 18 locales using the same contract.

EN· EnglishMSFT

§ 2 Bear case · Risk #1

AI capex overbuild — weak hyperscaler ROI

AI capex overbuild remains the leading downside risk. Microsoft's data-center spending stays elevated while AI monetization lags, and if hyperscaler ROI fails to materialize within 12–24 months, fair value compresses by an estimated 8–15%. Capex/Revenue at 22.9% and Capex/D&A at 1.89× already pressure free-cash-flow conversion below the five-year average.

FV impact: −8% to −15%
Timeline: 12–24 mo
Probability: Medium

Hold

DE· DeutschMSFT

§ 2 Bear-Case · Risiko Nr. 1

KI-Capex-Überbau — schwacher Hyperscaler-ROI

Der Überbau bei KI-Capex bleibt das wichtigste Abwärtsrisiko. Microsofts Rechenzentrums-Investitionen bleiben hoch, während die KI-Monetarisierung hinterherhinkt. Bleibt der Hyperscaler-ROI in den nächsten 12–24 Monaten aus, schrumpft der faire Wert um geschätzt 8–15%. Capex/Umsatz bei 22,9% und Capex/Abschreibungen bei 1,89× drücken die Free-Cash-Flow-Konversion bereits unter den Fünfjahresdurchschnitt.

FV impact: −8% to −15%
Timeline: 12–24 mo
Probability: Medium

Hold

JA· 日本語MSFT

§ 2 ベアケース · リスク #1

AI設備投資の過剰建設 — ハイパースケーラーROIの低迷

AI設備投資の過剰建設は最大の下振れリスクである。マイクロソフトのデータセンター投資は高水準のまま、AIの収益化が遅れている。今後12〜24か月でハイパースケーラーのROIが実現しない場合、フェアバリューは推定8〜15%圧縮される。売上高に対する設備投資比率22.9%、減価償却費に対する比率1.89倍は、すでにフリーキャッシュフロー転換率を5年平均以下に押し下げている。

FV impact: −8% to −15%
Timeline: 12–24 mo
Probability: Medium

Hold

AR· العربيةMSFTRTL

§ 2 الحالة الهبوطية · المخاطرة رقم 1

الإفراط في الإنفاق الرأسمالي على الذكاء الاصطناعي — ضعف العائد لمزودي السحابة

تظل مخاطر الإفراط في الإنفاق الرأسمالي على الذكاء الاصطناعي أبرز مخاطر التراجع. إذ يبقى إنفاق مايكروسوفت على مراكز البيانات مرتفعاً بينما يتأخر تحقيق العائد من الذكاء الاصطناعي، وفي حال لم يتحقق العائد على استثمار مزودي السحابة الكبار خلال 12–24 شهراً، فسينكمش العادل بنسبة تقديرية تتراوح بين 8% و15%. كما أن نسبة الإنفاق الرأسمالي إلى الإيرادات البالغة 22.9% ونسبتها إلى الاستهلاك البالغة 1.89 ضعفاً تضغطان بالفعل على تحويل التدفق النقدي الحر دون متوسط السنوات الخمس.

FV impact: −8% to −15%
Timeline: 12–24 mo
Probability: Medium

Hold

Languages

Methodology

Phase 0

Archetype detection

Each stock is classified into one of 9 archetypes. This calibrates every downstream parameter — discount rate, terminal growth, probability weights, scorecard weights, model applicability.

Hyper-growthNVDA → calibration parameters

Terminal growth band

Above default

High-growth tier band

Cost of equity band

Adjusted-beta band

CAPM at strict and moderate levels

Scorecard tilt

Growth + quality

Balance-sheet weight reduced for investment phase

Phase 1

Parallel data collection

Four research agents run concurrently: current ratios and analyst data; 5-year historical financials; peer multiples; WACC inputs and segments.

Parallel research agents

A1Current ratios + analyst data

Live ratios, prices, earnings summary, and forward analyst estimates with quality scores.

A2Standardised financials

Income statement, balance sheet, cash flow — annual + recent quarters, with multi-year trend context.

A3Peer multiples

Selected peers with forward P/E, EV/EBITDA, growth, margins. Used for PEG-adjusted relative value.

A4WACC + segments

Risk-free rate, industry beta, credit rating, segment KPIs, SBC dilution — feeding the cost-of-capital build.

Phase 2

Cost of capital

CAPM Ke at strict (raw beta) and moderate (adjusted-beta blend) levels. After-tax Kd. Market-weight WACC.

Ke strict

12.4%

Strict CAPM (raw beta)

Ke moderate

10.8%

Adjusted-beta blend

WACC

11.2%

Market-weight

Ke = Rf + β × ERP · computed at strict (raw beta) and moderate (adjusted-beta blend) levels · WACC market-weighted from Ke and after-tax Kd

Phase 2.5

Quality gates

ROIC vs WACC trajectory, accounting quality (OCF/NI, accruals, Beneish M-score). Gate failures increase required margin of safety.

Accounting quality gate

OCF / NIPass

Operating cash flow tracks reported earnings

Accruals ratioPass

Accrual share of earnings inside the healthy band

Beneish M-ScorePass

Composite manipulation screen clear of concern

ROIC vs WACCWide

Wide spread, multi-year trend favourable

Phase 3

Valuation models

SBC-adjusted discounted earnings (dual Ke), multi-stage moat fade, 5-year forward earnings (bull/base/bear), reverse DCF, Owner Earnings floor, PEG-adjusted peer.

Phase 3 output

5-year forward earnings — bull / base / bear

Bull scenario

Bull

Probability

25%

Return

+38%

Path target

$187

Data-center revenue compounds 32% / 3y
Gross margin holds 73%+
Software attach reaches 18% of revenue

Base scenario

Base

House call

Probability

55%

Return

+9%

Path target

$148

Data-center growth fades 32% → 18% by FY28
Gross margin compresses 200 bps
Capex intensity normalizes to 6% of revenue

Bear scenario

Bear

Probability

20%

Return

−24%

Path target

$103

Hyperscaler capex cycle tops Q3 2026
Custom-silicon takes 8 pts of accelerator share
Inventory write-down of one quarter of GPU stock

Phase 4

Sensitivity

5×5 Ke × terminal growth, TV concentration, growth × terminal P/E matrix.

Cost of equity

Terminal growth

	2.0%	2.5%	3.0%	3.5%	4.0%
10.0%	$172.00	$178.00	$185.00	$193.00	$202.00
11.0%	$156.00	$161.00	$167.00	$173.00	$180.00
12.0%	$142.00	$146.00	$151.00	$156.00	$162.00
13.0%	$129.00	$133.00	$137.00	$141.00	$146.00
14.0%	$118.00	$121.00	$124.00	$128.00	$132.00

5×5 grid of fair value vs Ke and terminal growth. Baseline cell outlined; tones grade upside vs current price.

Phase 5

Risk analysis — bear first

Pre-built kill scenarios, formal stress tests, quality deterioration signals. Built before bullish synthesis to counteract confirmation bias.

Phase 5 output

Top kill scenarios

Built before the bullish synthesis to counteract confirmation bias. Each scenario has a probability, fair-value impact, and trigger window.

Hyperscaler capex cycle peaks

30%· Medium

Top-4 cloud capex stops growing as ROIC on AI infra disappoints. Order book compresses 20–30% over two quarters before earnings revisions catch up.

FV impact

−22%

Trigger

12–18 mo

Custom-silicon share gain

35%· Med–High

TPU / Trainium / MTIA take share at the inference workload tier. NVDA accelerator unit growth slows even as data-center capex expands.

FV impact

−18%

Trigger

18–36 mo

Export-control escalation

20%· Low–Med

China license revocations or chip-design rule extensions remove 12–15% of revenue. Inventory write-down impairs near-term margin.

FV impact

−12%

Trigger

6–12 mo

Phase 6–7

Supplementary + actionable

Moat quantification, peer cohort, industry cycle, earnings decision tree, technical positioning, position sizing, dynamic margin of safety.

$135.40Price

FV $148

High $172

Owner-earnings floor$96.00

Moat

Widening ↗

8.1

/ 10

Wide moat

CUDA + ecosystem switching costs are widening, not eroding.

Switching costs (CUDA)Wide
Network effects (devs)Wide
Scale economiesModerate
BrandNarrow

Phase 8

Scorecard

9-category phase-adjusted score and 6-factor 100-point overlay with hard-fail guardrails.

Phase 8 output

9-category phase-adjusted scorecard

Business quality

9.2/10

Growth durability

8.4/10

Profitability

9.1/10

Balance sheet

8.6/10

Capital allocation

7.8/10

Moat durability

8.1/10

Management

8.4/10

Risk profile

6.4/10

Valuation

5.9/10

Weighted overall

7.9/10

Phase 9

Report assembly + 18-locale rendering

Assembled into a 500–900 line markdown document, translated on-view.

Final ratingAccumulateconf. 72/100

Composite fair value with explicit range and confidence.

Six-factor overlay surfaces the rating band only when both Risk and Valuation clear their hard-fail floors. Consensus divergence is checked against the diagnostic threshold before the rating is published.

Translated into 18 locales — same numbers everywhere

ENZHJAKODEFRESITPTNLPLTRRUARFAHITHVI