Posted on September 24, 2026September 24, 2026 | by rajeshkumar

Most investors begin stock research by asking a question such as:

“Is this company a good stock?”

That question sounds reasonable, but it is actually one of the worst ways to start investment research.

A company can be:

an outstanding business but an expensive stock,
a mediocre business at an extraordinary price,
a fast-growing company destroying shareholder value,
a statistically cheap company heading toward financial distress,
a turnaround before the financial statements show improvement,
or a fundamentally strong business whose stock price already discounts years of exceptional growth.

No single ratio—and no single AI model—can reliably capture all these dimensions.

The stronger approach is to build a multi-model stock research system in which independent analytical models examine different parts of the investment thesis and are combined only after their evidence has been reviewed.

Institutional factor investing follows a similar philosophy. MSCI, for example, identifies systematic equity factors including Value, Quality, Momentum, Low Size, Low Volatility and High Dividend Yield rather than relying on a single valuation ratio. MSCI also notes that multidimensional factors such as Value and Quality can be better captured through several underlying descriptors rather than one metric.

The objective of this framework is therefore not:

“Find stocks with the highest score.”

It is:

“Build a repeatable process capable of identifying exceptional businesses, mispriced securities, emerging compounders, credible turnarounds and asymmetric opportunities while systematically searching for evidence that could invalidate the thesis.”

PART I — THE CENTRAL IDEA

A serious stock-research system should separate at least nine questions:

1. What exactly does the company do?
2. Is the industry attractive?
3. Is the business competitively advantaged?
4. Are the financial statements healthy?
5. Is growth real, durable and value-creating?
6. Is management allocating capital intelligently?
7. What is the business intrinsically worth?
8. What expectations are already embedded in the share price?
9. What could permanently destroy the investment thesis?

These should not be answered by one giant subjective opinion.

Instead, construct independent research modules.

                           COMPANY
                              │
              ┌───────────────┼───────────────┐
              ↓               ↓               ↓
        BUSINESS MODEL    INDUSTRY MODEL    FINANCIAL MODEL
              │               │               │
              ↓               ↓               ↓
            MOAT             TAM           QUALITY
          PRICING          STRUCTURE          ROIC
          PRODUCT          DEMAND             FCF
          CUSTOMERS        CAPACITY           DEBT
              │               │               │
              └───────────────┼───────────────┘
                              ↓
                         GROWTH MODEL
                              │
                    Revenue / EPS / FCF
                    Reinvestment / ROIC
                              │
               ┌──────────────┼──────────────┐
               ↓              ↓              ↓
             DCF          RELATIVE       FACTOR MODEL
                         VALUATION
               │              │              │
               └──────────────┼──────────────┘
                              ↓
                       GOVERNANCE MODEL
                              │
                              ↓
                          RISK MODEL
                              │
                              ↓
                         RED-TEAM MODEL
                              │
                              ↓
                        FINAL RESEARCH
                            DOSSIER

Notice something important:

The final output is a research dossier—not simply a Buy/Hold/Sell button.

PART II — FIRST CLASSIFY THE COMPANY

This is one of the biggest improvements that can be made to conventional stock screeners.

You should not analyse every company using exactly the same model.

A 25%-growing electronics manufacturer should not be analysed like a steel producer.

A bank should not be valued like an EPC company.

A turnaround should not be rejected merely because its historical ROCE is poor.

Before calculating anything, classify the company.

Archetype	Main question
Compounder	Can high returns on capital persist while capital is reinvested?
Emerging growth company	How large can the business become and at what economics?
Turnaround	Is operational improvement genuine and sustainable?
Cyclical	Where are earnings relative to the economic cycle?
Commodity producer	What happens under normalized commodity prices?
Deep value	Are assets/cash flows worth materially more than the market price?
Financial company	Can book value compound at attractive risk-adjusted returns?
Asset-light platform	Can network effects/unit economics create durable profitability?
Capital-intensive growth	Will new capacity earn attractive incremental returns?
Event-driven/special situation	What catalyst can unlock value?

This single step prevents many analytical mistakes.

For example, Professor Aswath Damodaran emphasizes that cyclical and commodity businesses often require normalized earnings across the cycle rather than using whichever earnings happen to exist today.

Financial institutions require another approach entirely: debt is effectively part of their operating business, traditional enterprise-value concepts become less useful, and regulatory capital matters enormously. Damodaran therefore uses approaches such as dividend models, excess-return models and price-to-book comparisons when analysing banks and other financial institutions.

This leads to a crucial rule:

Classify first. Measure second.

PART III — MODEL 1: BUSINESS QUALITY

Before opening Excel, understand what generates the money.

A Business Model Score should examine six dimensions.

Dimension	Questions
Product	What exactly does the customer buy?
Customer	Who pays the company and why?
Revenue model	Recurring, transactional, project-based or cyclical?
Economics	Gross margin, contribution margin, capital intensity
Competitive position	Why can’t competitors easily take customers?
Reinforcement mechanism	Does getting larger make the business stronger?

A company deserves special attention when several desirable characteristics coexist:

Growing underlying market
+
Increasing market share
+
Pricing power
+
High incremental margins
+
High incremental ROIC
+
Long reinvestment runway

That combination is considerably more powerful than revenue growth alone.

PART IV — MODEL 2: INDUSTRY AND OPPORTUNITY

Excellent companies frequently emerge where several structural forces intersect.

Study:

Area	What to investigate
TAM	Current and realistic addressable market
Industry growth	Historical and expected demand
Penetration	Current adoption versus potential adoption
Supply	Capacity additions and competitive intensity
Regulation	Government incentives and restrictions
Import dependence	Potential localization opportunity
Export opportunity	International competitiveness
Technology	Risk of substitution or obsolescence
Customer economics	Whether adoption creates economic value
Industry structure	Fragmented versus concentrated

For Indian companies, particularly interesting structural themes may periodically include manufacturing localization, power infrastructure, electronics, defence, railways, data centres, renewable-energy infrastructure, digital financial infrastructure, specialty chemicals and industrial automation.

But a theme should never substitute for company analysis.

A booming industry can still contain terrible stocks.

PART V — MODEL 3: COMPETITIVE MOAT

Morningstar’s equity-research process explicitly considers sustainable competitive advantage alongside valuation and uncertainty. Its fair-value estimates use discounted cash-flow analysis, while its Economic Moat framework examines the durability of competitive advantage.

Common moat sources include:

Moat	Example characteristic
Network effect	Product becomes more valuable as users increase
Switching costs	Customers find changing vendors expensive
Cost advantage	Structural production/procurement advantage
Intangible assets	Brand, patents, licences, reputation
Efficient scale	Market cannot economically support many competitors
Distribution	Difficult-to-replicate channel network
Data advantage	More activity improves product/data
Ecosystem lock-in	Multiple interconnected products reinforce retention

But never award a moat merely because management says one exists.

Look for quantitative evidence.

Possible evidence:

ROIC persistently above cost of capital
Stable/high margins
Market-share gains
Low customer churn
High repeat revenue
Pricing power
Low customer acquisition cost
High asset turns
Increasing supplier/customer ecosystem

ROIC is particularly useful.

A simplified formulation is:

NOPAT = EBIT × (1 – normalized tax rate)

ROIC = NOPAT / Invested Capital

Economic value is created when returns on invested capital exceed the company’s cost of capital. Damodaran similarly describes economic value creation as the excess return earned over the cost of capital multiplied by invested capital.

Therefore:

ROIC > WACC

is substantially more meaningful than:

Revenue is growing quickly.

PART VI — MODEL 4: FINANCIAL QUALITY

Growth becomes much more valuable when accompanied by financial quality.

The Quality Model should examine four broad areas.

Profitability

Gross Margin
EBITDA Margin
EBIT Margin
Net Margin
ROA
ROE
ROIC

Balance-sheet strength

Net Debt / EBITDA
Debt / Equity
Interest Coverage
Working Capital
Current Ratio
Cash Balance
Debt maturity profile

Cash generation

CFO
Free Cash Flow
CFO / PAT
FCF / PAT
FCF / EBITDA
Cash ROIC

Stability

Study five- to ten-year behaviour where possible:

Revenue volatility
Margin volatility
EPS volatility
ROIC volatility
Cash-flow volatility

MSCI’s Quality methodology, for example, incorporates profitability, leverage and earnings stability using descriptors including ROE, debt/equity and earnings variability.

PART VII — MODEL 5: EARNINGS QUALITY

Reported accounting profit is not equivalent to economic cash generation.

A company may report rapidly increasing PAT while simultaneously experiencing deteriorating cash conversion.

Therefore compare:

PAT
vs.
Cash Flow From Operations
vs.
Free Cash Flow

A useful diagnostic is:

CFO / PAT

Investigate persistent discrepancies.

Forensic checks should include:

Area	Warning signal
Receivables	Growing materially faster than sales
Inventory	Rising despite weak revenue
Advances	Large unexplained advances
Other income	Supporting reported profit
Capitalization	Expenses being capitalized aggressively
Working capital	Consuming increasing amounts of cash
Tax	Persistently abnormal effective tax rates
Subsidiaries	Complex or opaque structures
Auditor	Resignations or repeated qualifications
Contingent liabilities	Large relative to net worth
Related parties	Large or unusual transactions
Dilution	Repeated equity issuance
Cash vs debt	Large cash balance alongside expensive borrowing

A company with:

PAT CAGR = 30%
CFO CAGR = 4%
Receivable CAGR = 45%

requires substantially more investigation than the headline 30% profit growth suggests.

PART VIII — PIOTROSKI F-SCORE

Joseph Piotroski’s F-Score was introduced in his 2000 research on separating stronger from weaker high-book-to-market companies. It combines nine accounting signals covering profitability, financial structure/liquidity and operating efficiency.

The nine tests are broadly:

Category	Test
Profitability	Positive ROA
Profitability	Positive operating cash flow
Profitability	Improving ROA
Profitability	CFO exceeds accounting earnings
Financial structure	Lower leverage
Liquidity	Improving current ratio
Financing	No new equity issuance
Efficiency	Improving gross margin
Efficiency	Improving asset turnover

Each test receives:

Pass = 1
Fail = 0

giving:

F-Score = 0 to 9

Piotroski originally used this as a way of separating stronger and weaker companies within a value universe—not as a universal stock-selection system.

That limitation matters.

A company can have:

Piotroski = 9

and still be:

Overvalued
Structurally declining
Poorly governed
Cyclically near peak earnings

Therefore:

Use Piotroski as a financial-health diagnostic, not a Buy signal.

PART IX — ALTMAN Z-SCORE

Edward Altman’s original 1968 research combined multiple financial ratios to estimate corporate bankruptcy risk.

The classic publicly traded manufacturing-company model is:

Z =
1.2 × X1
+ 1.4 × X2
+ 3.3 × X3
+ 0.6 × X4
+ 1.0 × X5

where:

X1 = Working Capital / Total Assets
X2 = Retained Earnings / Total Assets
X3 = EBIT / Total Assets
X4 = Market Value of Equity / Total Liabilities
X5 = Sales / Total Assets

The important lesson is not simply the resulting number.

The five variables force you to examine:

Liquidity
Accumulated profitability
Operating productivity
Market-value cushion
Asset efficiency

Do not blindly apply the original Altman model across banks, financial companies or businesses whose economics differ substantially from the manufacturing companies for which the original model was developed.

PART X — MODEL 6: GROWTH

Growth should be decomposed.

Do not stop at:

Revenue CAGR = 25%

Calculate:

Revenue CAGR
EBITDA CAGR
EBIT CAGR
PAT CAGR
EPS CAGR
CFO CAGR
FCF CAGR

The CAGR formula is:

CAGR = (Ending Value / Beginning Value)^(1 / Years) – 1

Then determine where growth actually comes from.

Volume growth
+
Price increases
+
New products
+
New customers
+
Geographic expansion
+
Acquisitions
+
New capacity
+
Market-share gains

Organic growth deserves different treatment from acquisition-funded growth.

PART XI — INCREMENTAL RETURNS ARE MORE IMPORTANT THAN HISTORICAL RETURNS

Suppose a company historically generated:

ROIC = 28%

but its new factories produce:

Incremental ROIC = 11%

The headline historical ROIC may be misleading.

Calculate:

Incremental ROIC
=
Change in NOPAT
/
Change in Invested Capital

A particularly powerful compounder can often be described as:

High ROIC
×
High Reinvestment Rate
×
Long Reinvestment Runway

This is the economic engine behind many long-term wealth creators.

Growth itself does not create value.

Profitable reinvestment creates value.

PART XII — MODEL 7: DCF VALUATION

Discounted Cash Flow attempts to answer:

What is the present value of the future cash flows generated by the business?

For a standard FCFF model:

Enterprise Value
=
Σ FCFF(t)/(1 + WACC)^t
+
Terminal Value/(1 + WACC)^n

A perpetual-growth terminal value is commonly expressed as:

TV =
FCFF(n+1)
/
(WACC – g)

Damodaran notes that terminal value represents the cash flows beyond the explicit forecast period and that using a market multiple for terminal value mixes relative valuation with intrinsic valuation.

From enterprise value:

Equity Value
=
Enterprise Value
+ Cash
+ Non-operating investments
– Debt
– Other senior claims

and:

Intrinsic Value Per Share
=
Equity Value
/
Diluted Shares Outstanding

PART XIII — NEVER USE ONE DCF

DCF creates a dangerous illusion of precision.

Suppose your spreadsheet reports:

Fair Value = ₹1,842.37

The .37 is meaningless if tiny changes in WACC or long-term margins move the valuation by hundreds of rupees.

Use scenarios.

Variable	Bear	Base	Bull
Revenue CAGR	Lower	Central	Higher
Terminal margin	Lower	Central	Higher
WACC	Higher	Central	Lower
Terminal growth	Lower	Central	Higher
Fair value	X	Y	Z

Then calculate sensitivity tables for:

WACC
vs.
Terminal Growth

and:

Revenue growth
vs.
Operating margin

The output should therefore be:

Valuation range

rather than:

Exact fair value

Morningstar similarly combines its DCF-based fair-value estimates with an explicit assessment of uncertainty rather than treating valuation as perfectly precise.

PART XIV — REVERSE DCF

One of the strongest additions to conventional valuation is the Reverse DCF.

Traditional DCF asks:

“What should the stock be worth based on my assumptions?”

Reverse DCF asks:

“What assumptions must be true to justify today’s stock price?”

Suppose the market valuation requires:

25% revenue CAGR for 10 years
+
22% operating margin
+
High terminal growth

while competitors earn:

12% operating margins

The important observation may not be that your DCF is “right.”

It may be that the stock price embeds unusually optimistic assumptions.

Reverse DCF converts valuation analysis into an expectations test.

PART XV — RELATIVE VALUATION

DCF should be complemented with relative valuation.

Common measures include:

P/E
EV/EBITDA
EV/EBIT
P/B
P/S
FCF Yield
Earnings Yield
PEG

But comparisons must be economically sensible.

Do not compare:

A bank
with
a software business

merely because both trade publicly.

Compare companies with similar:

Industry
Business model
Growth
ROIC
Margins
Capital intensity
Balance-sheet risk

MSCI’s research similarly emphasizes that multi-descriptor approaches can improve factor definitions and reduce problems arising from relying on one measure alone.

PART XVI — THE VALUATION TRIANGLE

A strong valuation process uses three different perspectives:

                  INTRINSIC
                     DCF
                      ▲
                     / \
                    /   \
                   /     \
                  /       \
                 ▼─────────▼
          RELATIVE       HISTORY
         Peer multiples   Own historical
                          valuation range

Ask three questions:

1. What is the business intrinsically worth?
2. How is it priced relative to comparable companies?
3. Where is it trading relative to its own historical valuation?

When all three point in a similar direction, confidence increases.

When they strongly disagree, investigate why.

PART XVII — MODEL 8: FACTOR ANALYSIS

Factor analysis answers a different question from valuation.

DCF asks:

What is the company worth?

Factor analysis asks:

What characteristics does this security have relative to other securities?

A practical factor engine can include:

VALUE
QUALITY
GROWTH
MOMENTUM
LOW VOLATILITY
SIZE
YIELD

Growth can be extremely useful in a research system even though institutional factor classifications differ on whether to treat it as a standalone persistent factor.

PART XVIII — VALUE FACTOR

Possible descriptors:

Earnings Yield
FCF Yield
EBIT / EV
Book / Price
Sales / EV
Normalized Earnings Yield

Avoid using only P/E.

P/E becomes meaningless or misleading when:

Earnings are negative
Earnings are temporarily depressed
Earnings are cyclically inflated
Accounting profits do not convert to cash

PART XIX — QUALITY FACTOR

Possible descriptors:

ROIC
ROE
ROA
Gross margin
Operating margin
CFO / PAT
FCF conversion
Debt / Equity
Interest coverage
Earnings stability

Ideally quality should reward businesses demonstrating:

High profitability
+
Cash-backed earnings
+
Low financial risk
+
Stable economics

PART XX — GROWTH FACTOR

Possible inputs:

3Y Revenue CAGR
5Y Revenue CAGR
3Y EBITDA CAGR
3Y EPS CAGR
3Y CFO CAGR
3Y FCF CAGR
Forward growth
Order-book growth
Capacity growth

A better growth score also examines:

Growth consistency
Growth quality
Incremental ROIC
Organic versus acquired growth

PART XXI — MOMENTUM FACTOR

Momentum should not replace fundamental research.

It answers:

Is the market increasingly or decreasingly agreeing with the thesis?

Possible measurements:

3-month return
6-month return
12-month return
Return relative to Nifty 500
Return relative to sector index
Distance from 52-week high
Earnings-revision momentum
Volume trend

MSCI’s momentum framework combines recent six- and twelve-month price performance and risk-adjusts the resulting momentum measure.

Momentum is particularly useful for detecting situations where:

Fundamentals improve
+
Earnings estimates rise
+
Price strength emerges

That combination can indicate a genuine operating inflection.

PART XXII — DO NOT SCORE RAW METRICS

Imagine comparing:

Company A ROIC = 31%
Company B ROIC = 22%
Company C ROIC = 17%

and:

Company A P/E = 70
Company B P/E = 24
Company C P/E = 12

You cannot meaningfully average:

ROIC + P/E + CAGR

because their units and distributions differ.

Normalize first.

A simple method is percentile ranking.

Metric Score
=
Percentile rank among comparable companies × 100

For metrics where lower is desirable, reverse the percentile.

For example:

Higher ROIC        → better
Higher FCF yield   → better
Lower EV/EBITDA    → better
Lower leverage     → better
Lower volatility   → better

Most importantly:

Normalize within economically comparable sectors.

A 15% ROE means something very different for a regulated utility, bank, software company and steel producer.

PART XXIII — REMOVE OUTLIERS

Financial datasets contain extreme observations.

Before normalization, consider winsorizing extreme values.

For example:

2.5th percentile → floor
97.5th percentile → ceiling

This prevents one strange observation such as:

P/E = 1,800×

from distorting the entire model.

Robust systems can also use:

Median
Median Absolute Deviation
Percentile ranks

instead of simple means and standard deviations.

PART XXIV — A PRACTICAL FACTOR SCORE

An example factor model might be:

Factor Score =
25% Quality
+ 20% Value
+ 20% Growth
+ 20% Momentum
+ 10% Low Volatility
+ 5% Size/Liquidity adjustment

The exact weights are less important than ensuring that:

Metrics are defined consistently
Data is point-in-time
Metrics aren't heavily duplicated
Industry biases are controlled
Outliers are controlled

Do not optimize weights until your historical backtest becomes beautiful.

That is one of the easiest ways to build an overfitted model.

PART XXV — MODEL 9: MANAGEMENT AND GOVERNANCE

Numbers tell you what happened.

Management analysis helps determine why it happened and whether it can continue.

Evaluate:

Area	Questions
Capital allocation	Where does management invest retained earnings?
Acquisitions	Have acquisitions created shareholder value?
Dilution	Is equity repeatedly issued?
Dividends/buybacks	Are distributions economically sensible?
Guidance	Does management consistently deliver what it promises?
Compensation	Are incentives aligned with shareholders?
Related parties	Are transactions transparent and reasonable?
Promoter ownership	Is ownership stable?
Pledging	Are promoter shares pledged?
Communication	Are problems acknowledged openly?

For Indian listed companies, NSE makes corporate announcements, financial results and shareholding information available through its corporate-filings system.

NSE also maintains promoter-pledge/encumbrance data, providing a valuable governance and financing-risk input.

SEBI’s disclosure framework also covers related-party transactions, making RPT disclosures an important primary source during governance research.

PART XXVI — MANAGEMENT PROMISE VS DELIVERY

Create a historical table.

Year	Management statement	Actual outcome	Result
FY22	Capacity +30%	+31%	Delivered
FY23	Margin 15–17%	12%	Missed
FY24	Debt reduction	Debt increased	Missed
FY25	₹2,000 cr revenue	₹2,060 cr	Delivered

Do this for five years if enough history exists.

It is one of the simplest ways to separate:

Excellent storytellers

from:

Excellent operators.

PART XXVII — MODEL 10: RISK

Risk should not mean merely share-price volatility.

Separate risk into categories.

BUSINESS RISK
Customer concentration
Supplier dependence
Product concentration
Technology disruption

FINANCIAL RISK
Debt
Interest coverage
Working-capital requirement
Refinancing

GOVERNANCE RISK
Related parties
Pledging
Auditor issues
Dilution

INDUSTRY RISK
Competition
Regulation
Commodity cycle
Overcapacity

VALUATION RISK
Optimistic expectations
Multiple compression
DCF sensitivity

MARKET RISK
Liquidity
Volatility
Beta
Drawdown

The most dangerous investment risks are often permanent-loss risks, not temporary price fluctuations.

PART XXVIII — HARD RISK GATES

This is another important improvement over simply averaging everything.

Suppose:

Business = 92
Growth = 95
Valuation = 72
Momentum = 90

but:

Auditor resigned
Promoter pledged most holdings
Large unexplained related-party transactions exist

A simple weighted average could still produce:

Final Score = 84

That would be absurd.

Some risks should act as gates, not small negative scores.

Examples requiring manual investigation:

Auditor resignation
Serious qualified audit opinion
Material regulatory investigation
Large unexplained RPT
Rapid unexplained receivable growth
Severe liquidity pressure
Repeated promoter pledging
Frequent equity dilution
Major contingent liability
Accounting restatement

The stock can remain under research, but the system should prevent a high composite score from hiding a potentially existential issue.

PART XXIX — BUILD A RED-TEAM MODEL

This may be the most valuable model in the entire framework.

Instead of asking:

“Why should I invest?”

assign one analyst—or one AI—to answer:

“Build the strongest possible evidence-based case for why this investment thesis could fail.”

The red team investigates:

Overstated TAM
Unsustainable margins
Customer concentration
Technology disruption
Promoter risk
Aggressive accounting
Capacity glut
Commodity-cycle peak
Government-policy dependence
Unrealistic valuation assumptions
Competitor response
Execution risk

The red team’s purpose is not pessimism.

Its purpose is to fight confirmation bias.

PART XXX — DON’T DOUBLE COUNT INFORMATION

This is where many sophisticated-looking scorecards fail.

Suppose your model contains:

ROE
ROIC
Operating Margin
Net Margin
Quality Factor
Piotroski

and then gives each one equal weight.

You may believe six independent signals exist.

In reality, several may be measuring largely the same underlying profitability phenomenon.

Similarly:

P/E
Earnings Yield
PEG

are highly related.

And:

Debt/Equity
Debt/EBITDA
Interest Coverage
Altman Z

partly overlap.

Therefore group correlated metrics into modules before assigning weights.

PART XXXI — THE GOLD-STANDARD SCORECARD

A practical architecture might be:

Module	Weight
Business & Industry	20%
Financial Quality	20%
Growth & Reinvestment	15%
Valuation	20%
Management & Governance	10%
Factor/Market Confirmation	5%
Risk Resilience	10%
Total	100%

But first apply:

DATA QUALITY GATE
        ↓
ACCOUNTING/GOVERNANCE GATE
        ↓
SOLVENCY GATE
        ↓
LIQUIDITY GATE
        ↓
100-POINT RESEARCH MODEL

This prevents a mathematically attractive score from masking a fundamental problem.

PART XXXII — ADD A CONFIDENCE SCORE

This is extremely important.

A company may receive:

Research Score = 86/100

but the analysis may depend on:

Only two years of history
Opaque subsidiaries
Uncertain margins
No reliable management guidance
New business model

Therefore publish a separate:

CONFIDENCE SCORE

Example:

Research Score:   86/100
Confidence:       58/100

These mean completely different things.

Research Score asks:

How attractive are the observed characteristics?

Confidence asks:

How reliable is the evidence supporting that conclusion?

Never combine the two.

PART XXXIII — SECTOR-SPECIFIC MODELS

A universal model should contain specialized modules.

Banks

Focus more heavily on:

ROA
ROE
Loan growth
Deposit growth
CASA
NIM
GNPA
NNPA
Provision coverage
Credit cost
Capital adequacy
Book-value growth
P/B
P/E

Traditional industrial-company EV/EBITDA analysis is often inappropriate because debt is effectively part of the operating business of a financial institution.

NBFCs

Add:

Borrowing cost
ALM mismatch
AUM growth
Credit quality
Capital adequacy
Funding diversification

Insurance

Study:

Embedded Value
VNB
VNB margin
APE growth
Persistency
Solvency
P/EV

EPC / Infrastructure

Focus on:

Order book
Order inflow
Execution
Book-to-bill
Receivable days
Working capital
Operating cash flow
ROCE
Debt
Customer quality

Manufacturing

Study:

Capacity
Utilization
Realization
Volume growth
Asset turns
Incremental ROCE
Raw-material sensitivity
Export mix

Commodity Businesses

Normalize earnings across the cycle.

Damodaran suggests using long-term averages, normalized margins or sector-cycle averages when analysing cyclical and commodity companies.

High-Growth Technology / Platforms

Focus more heavily on:

Users/customers
Retention
ARPU
Gross margin
Contribution margin
CAC
Payback
Unit economics
Market share
Operating leverage

High-growth businesses can have little current profit despite significant potential value, which makes conventional current P/E comparisons particularly weak.

PART XXXIV — FINDING EMERGING OR “HIDDEN” OPPORTUNITIES

Do not begin by asking:

“Which stocks will become multibaggers?”

No model can reliably know that in advance.

Instead search for conditions frequently associated with major fundamental re-ratings.

One useful discovery equation is:

Structural Industry Growth
×
Market Share Gains
×
Operating Leverage
×
Capacity Expansion
×
Improving ROIC
×
Improving Cash Flow
×
Balance-Sheet Strength
×
Reasonable Starting Valuation

Particularly interesting situations often occur when several changes happen simultaneously:

Revenue acceleration
+
Margin expansion
+
Debt reduction
+
Capacity coming online
+
New customer wins
+
Industry tailwind

The key concept is:

INFLECTION

The market often already understands companies that are excellent.

What can be more interesting is a company where the rate of improvement itself is changing.

PART XXXV — TURNAROUND MODEL

A turnaround should have its own framework.

Look for sequential evidence.

Stage 1 — Industry environment stops deteriorating
Stage 2 — Revenue decline stabilizes
Stage 3 — Gross margin improves
Stage 4 — EBITDA recovers
Stage 5 — Cash flow improves
Stage 6 — Debt falls
Stage 7 — ROCE improves
Stage 8 — Market recognizes recovery

An early turnaround may therefore look statistically unattractive:

Low ROE
Low margins
High P/E
Weak trailing FCF

because trailing financial statements describe the old business condition.

This is why historical quality screens alone can miss turnarounds.

PART XXXVI — THE DATA HIERARCHY

Not all sources deserve equal trust.

Use approximately this order:

LEVEL 1
Exchange filings
Annual reports
Audited financial statements
SEBI/RBI/regulatory documents

LEVEL 2
Quarterly results
Investor presentations
Credit-rating reports
Concall transcripts

LEVEL 3
Industry associations
Competitor filings
Government statistics

LEVEL 4
Financial databases/screeners

LEVEL 5
Broker reports
News articles

LEVEL 6
Social media
YouTube
Forums

Screeners are excellent for discovering ideas.

They should not be the final authority for investment-critical facts.

For Indian listed companies, NSE’s disclosure system itself notes that listed companies submit fundamentals, corporate announcements and shareholding information through exchange filings.

PART XXXVII — BUILD A FACT SHEET BEFORE FORMING AN OPINION

Before deciding whether you like a company, construct a standardized historical dataset.

At minimum collect five to ten years where available.

Revenue
EBITDA
EBIT
PAT
EPS

CFO
Capex
FCF

Assets
Equity
Debt
Cash

ROE
ROCE
ROIC

Margins

Receivables
Inventory
Payables

Shares outstanding
Promoter holding
Promoter pledge

Major acquisitions
Major capex
Dividend/buyback history

Only after building the fact sheet should qualitative conclusions begin.

This dramatically reduces narrative bias.

PART XXXVIII — THE MULTI-AI RESEARCH ARCHITECTURE

Now comes the most powerful implementation.

Instead of asking one AI:

“Analyse XYZ stock.”

create an AI investment committee.

                         COMPANY
                            │
          ┌─────────────────┼─────────────────┐
          ↓                 ↓                 ↓
     SOURCE AGENT     BUSINESS AGENT    INDUSTRY AGENT
          │                 │                 │
          └─────────────────┼─────────────────┘
                            ↓
                    FINANCIAL FORENSICS
                            │
             ┌──────────────┼──────────────┐
             ↓              ↓              ↓
       VALUATION AGENT  FACTOR AGENT   GOVERNANCE AGENT
             │              │              │
             └──────────────┼──────────────┘
                            ↓
                       RED-TEAM AGENT
                            │
                            ↓
                      SYNTHESIS AGENT
                            │
                            ↓
                      RESEARCH DOSSIER

PART XXXIX — AGENT RESPONSIBILITIES

Source Agent

Collect facts only.

No investment opinion.

It should gather:

Annual reports
Quarterly filings
Shareholding
Pledge data
Investor presentations
Concall transcripts
Credit ratings
Regulatory disclosures
Industry statistics

Every number should include:

Source
Period
Date

Business Analyst

Analyse:

Business model
Products
Customers
Competitive advantage
Pricing power
Market structure
TAM

Financial Forensics Agent

Analyse:

Revenue
Margins
ROIC
Cash conversion
Working capital
Debt
Piotroski
Altman where appropriate
Accounting anomalies

Valuation Agent

Build:

DCF
Reverse DCF
Peer valuation
Historical valuation
Bear/Base/Bull scenarios

Quant Agent

Calculate:

Value
Quality
Growth
Momentum
Volatility
Relative strength

Governance Agent

Analyse:

Promoter holding
Pledging
Dilution
Related parties
Auditor
Capital allocation
Management promises

Red-Team Agent

It receives the bullish thesis and tries to destroy it with evidence.

Synthesis Agent

Only this agent sees all conclusions.

It identifies:

Agreements
Contradictions
Missing evidence
Critical assumptions
Thesis-breaking risks

PART XL — WHY THE AI AGENTS SHOULD WORK INDEPENDENTLY

Do not show Agent B the conclusion from Agent A before B performs its own research.

Otherwise you create AI confirmation bias.

A better process is:

Independent analysis
        ↓
Independent analysis
        ↓
Independent analysis
        ↓
Compare results
        ↓
Investigate disagreement

Disagreement is useful.

Suppose:

Business Agent:     Excellent
Financial Agent:    Excellent
Valuation Agent:    Expensive
Red Team:           Customer concentration dangerous

That disagreement contains more information than five AIs all saying:

“Strong company.”

PART XLI — MASTER AI PROMPT

A reusable master prompt could look like this:

Act as an institutional-quality equity research team.

Company: [COMPANY]
Ticker: [TICKER]
Exchange: [NSE/BSE]
Research Date: [DATE]

Do not begin with a Buy/Sell conclusion.

Use current primary sources whenever possible.

Separate facts, calculations, assumptions and opinions.

Research the company through independent modules:

1. Business model
2. Industry/TAM
3. Competitive moat
4. Historical financial performance
5. Earnings quality
6. ROIC and reinvestment
7. Growth durability
8. Balance-sheet risk
9. Piotroski F-Score where applicable
10. Altman Z-Score where applicable
11. DCF valuation
12. Reverse DCF
13. Peer valuation
14. Historical valuation
15. Factor analysis
16. Management and governance
17. Capital allocation
18. Promoter/shareholding analysis
19. Risk analysis
20. Red-team analysis

For every important claim provide:

- source
- date
- evidence
- calculation where applicable

Do not invent missing financial data.

Do not treat management guidance as fact.

Identify contradictions between management claims and historical results.

Create Bear, Base and Bull operating scenarios.

Separate:

Research Score
Valuation Range
Confidence Score

Do not allow serious governance, solvency or accounting risks
to disappear inside a weighted average.

Conclude with:

A. One-sentence business description
B. Investment thesis
C. Evidence supporting thesis
D. Evidence against thesis
E. Five key operating KPIs
F. Five thesis-breaking risks
G. Valuation range
H. Market expectations implied by current price
I. Upcoming catalysts
J. Conditions that would invalidate the thesis
K. Questions still unanswered

That prompt is dramatically more powerful than:

Should I buy this stock?

PART XLII — THE FINAL COMPANY RESEARCH CARD

Every company should eventually fit onto one summary card.

COMPANY: ______________________
DATE: _________________________
PRICE: ________________________

ARCHETYPE
Compounder / Growth / Turnaround /
Cyclical / Financial / Deep Value

BUSINESS QUALITY
__/100

FINANCIAL QUALITY
__/100

GROWTH & REINVESTMENT
__/100

VALUATION
__/100

MANAGEMENT/GOVERNANCE
__/100

FACTOR/MARKET
__/100

RISK RESILIENCE
__/100

────────────────────────────

RESEARCH SCORE
__/100

CONFIDENCE SCORE
__/100

────────────────────────────

DCF RANGE
Bear: ₹____
Base: ₹____
Bull: ₹____

CURRENT PRICE
₹____

────────────────────────────

5 MOST IMPORTANT KPIs

1.
2.
3.
4.
5.

────────────────────────────

THESIS BREAKERS

1.
2.
3.
4.
5.

────────────────────────────

NEXT RESULTS TO WATCH
___________________________

NEXT CATALYST
___________________________

PART XLIII — MONITOR THE THESIS, NOT THE SHARE PRICE

Once a company enters the portfolio or watchlist, define what matters.

For example, for an EPC company:

Order inflow
Execution growth
EBITDA margin
Receivable days
Operating cash flow

For an electronics manufacturer:

Capacity utilization
Customer additions
Revenue growth
Gross margin
ROCE

For a broker:

Active clients
Market share
ARPU
Revenue diversification
Regulatory changes

The question every quarter becomes:

Is the underlying thesis strengthening, weakening or unchanged?

rather than:

“The share fell 8%. Should I sell?”

PART XLIV — BUILD A THESIS-INVALIDATION SYSTEM

Before investing, explicitly state:

I will reconsider this thesis if:

and define measurable conditions.

For example:

Revenue growth <10% for two consecutive quarters

OR

EBITDA margin falls below 12%

OR

Net Debt/EBITDA exceeds 2×

OR

Promoter pledge materially increases

OR

Major customer's contribution falls unexpectedly

OR

Capacity ramp is delayed >12 months

This converts emotional decision-making into evidence-based monitoring.

PART XLV — BACKTESTING THE MODEL

Before trusting a quantitative score, test it historically.

But avoid common backtesting errors:

Survivorship bias
Look-ahead bias
Using restated financial data
Ignoring delisted companies
Ignoring transaction costs
Ignoring liquidity
Overfitting factor weights
Data-mining hundreds of metrics

Most importantly, use point-in-time information.

If analysing what the model would have selected on June 30, 2021, it should only use information an investor could actually have known on June 30, 2021.

Anything else creates artificial performance.

PART XLVI — MULTIBAGGER RESEARCH IS NOT ABOUT PREDICTING PRICE

The phrase “multibagger” can tempt investors into searching for stocks capable of going up five or ten times.

A better framework asks:

Can earnings become 3×–5× larger?

Then:

Can the company maintain or improve returns on capital?

Then:

How much of that growth does today's valuation already assume?

A sustainable multibagger generally requires some combination of:

Earnings growth
+
Cash-flow growth
+
ROIC sustainability
+
Long reinvestment runway
+
Possibly valuation re-rating

The safest foundation is usually fundamental compounding, not hoping the P/E multiple becomes permanently larger.

PART XLVII — THE MOST INTERESTING SIGNAL: MULTIPLE INFLECTION

Some of the best research candidates occur when several independent variables improve together.

For example:

Revenue Growth        ↑
EBITDA Margin         ↑
ROIC                  ↑
Operating Cash Flow   ↑
Debt                  ↓
Order Book            ↑
Capacity Utilization  ↑
Market Share          ↑

One improving number may be noise.

Six independently improving variables can represent a genuine business inflection.

This is where a multi-model framework becomes especially useful.

PART XLVIII — WHAT NOT TO DO

Avoid building a system that says:

P/E good       +1
ROE good       +1
Revenue good   +1
Momentum good  +1
DCF good       +1

5/5 = BUY

That is not serious research.

The goal is to understand:

BUSINESS
        ↓
ECONOMIC ENGINE
        ↓
FINANCIAL EVIDENCE
        ↓
FUTURE OPPORTUNITY
        ↓
MANAGEMENT
        ↓
VALUATION
        ↓
RISKS
        ↓
MARKET EXPECTATIONS

Only then should the investment thesis be written.

PART XLIX — THE COMPLETE RESEARCH PIPELINE

The entire process can finally be expressed as:

                    STOCK UNIVERSE
                          │
                          ↓
                 DATA QUALITY FILTER
                          │
                          ↓
                  COMPANY ARCHETYPE
                          │
                          ↓
                  INDUSTRY ANALYSIS
                          │
                          ↓
                  BUSINESS QUALITY
                          │
                          ↓
                       MOAT
                          │
                          ↓
                 FINANCIAL QUALITY
                          │
                          ↓
                  EARNINGS QUALITY
                          │
                          ↓
               GROWTH + REINVESTMENT
                          │
                          ↓
                     ROIC / WACC
                          │
                          ↓
               ┌──────────┼──────────┐
               ↓          ↓          ↓
              DCF      RELATIVE    REVERSE
                       VALUATION      DCF
               │          │          │
               └──────────┼──────────┘
                          ↓
                     FACTOR MODEL
                          │
                          ↓
                       MOMENTUM
                          │
                          ↓
                     MANAGEMENT
                          │
                          ↓
                     GOVERNANCE
                          │
                          ↓
                        RISK
                          │
                          ↓
                      RED TEAM
                          │
                          ↓
                HARD-RISK GATE
                          │
                          ↓
                   RESEARCH SCORE
                          +
                   CONFIDENCE SCORE
                          +
                    VALUATION RANGE
                          │
                          ↓
                   RESEARCH DOSSIER
                          │
                          ↓
                    WATCHLIST /
                     MONITORING

FINAL PRINCIPLE

The biggest mistake in stock research is searching for one magical metric.

There isn’t one.

DCF can be wrong because forecasts are wrong.

P/E can be misleading because earnings are cyclical.

ROE can be artificially boosted by leverage.

Revenue growth can destroy value.

Momentum can reverse.

Piotroski can identify financial strength without identifying valuation.

Altman can identify financial distress without telling you whether the business is attractive.

A moat can exist while the stock remains wildly overpriced.

This is precisely why the strongest research process uses multiple independent lenses.

The goal is not to find a stock where every model agrees.

The goal is to understand why the models disagree.

That disagreement often contains the most valuable information.

The resulting research system can be summarized as:

FUNDAMENTALS
+
BUSINESS QUALITY
+
INDUSTRY STRUCTURE
+
MOAT
+
FINANCIAL QUALITY
+
EARNINGS QUALITY
+
ROIC / REINVESTMENT
+
GROWTH
+
DCF
+
REVERSE DCF
+
RELATIVE VALUATION
+
FACTOR ANALYSIS
+
MOMENTUM
+
MANAGEMENT
+
GOVERNANCE
+
RISK
+
RED TEAM
+
THESIS MONITORING
────────────────────────────
REPEATABLE INVESTMENT RESEARCH

That is the difference between asking:

“Do you like this stock?”

and building a system capable of asking:

“What does this business actually deserve to be worth, what assumptions does today’s market price require, what evidence supports those assumptions, what evidence contradicts them, and what would have to happen for my thesis to be wrong?”

That is a much better question.

And answering that question repeatedly, consistently and without changing the rules depending on which company you already want to own is what turns stock research from an opinion into a research process.

MOTOSHARE 🚗🏍️

Turn Idle Vehicles into Income

How to Build a Repeatable Multi-Model System for Fundamental Analysis, Valuation, Quality, Growth, Momentum, Risk and Governance

PART I — THE CENTRAL IDEA

PART II — FIRST CLASSIFY THE COMPANY

PART III — MODEL 1: BUSINESS QUALITY

PART IV — MODEL 2: INDUSTRY AND OPPORTUNITY

PART V — MODEL 3: COMPETITIVE MOAT

PART VI — MODEL 4: FINANCIAL QUALITY

Profitability

Balance-sheet strength

Cash generation

Stability

PART VII — MODEL 5: EARNINGS QUALITY

PART VIII — PIOTROSKI F-SCORE

PART IX — ALTMAN Z-SCORE

PART X — MODEL 6: GROWTH

PART XI — INCREMENTAL RETURNS ARE MORE IMPORTANT THAN HISTORICAL RETURNS

PART XII — MODEL 7: DCF VALUATION

PART XIII — NEVER USE ONE DCF

PART XIV — REVERSE DCF

PART XV — RELATIVE VALUATION

PART XVI — THE VALUATION TRIANGLE

PART XVII — MODEL 8: FACTOR ANALYSIS

PART XVIII — VALUE FACTOR

PART XIX — QUALITY FACTOR

PART XX — GROWTH FACTOR

PART XXI — MOMENTUM FACTOR

PART XXII — DO NOT SCORE RAW METRICS

PART XXIII — REMOVE OUTLIERS

PART XXIV — A PRACTICAL FACTOR SCORE

PART XXV — MODEL 9: MANAGEMENT AND GOVERNANCE

PART XXVI — MANAGEMENT PROMISE VS DELIVERY

PART XXVII — MODEL 10: RISK

PART XXVIII — HARD RISK GATES

PART XXIX — BUILD A RED-TEAM MODEL

PART XXX — DON’T DOUBLE COUNT INFORMATION

PART XXXI — THE GOLD-STANDARD SCORECARD

PART XXXII — ADD A CONFIDENCE SCORE

PART XXXIII — SECTOR-SPECIFIC MODELS

Banks

NBFCs

Insurance

EPC / Infrastructure

Manufacturing

Commodity Businesses

High-Growth Technology / Platforms

PART XXXIV — FINDING EMERGING OR “HIDDEN” OPPORTUNITIES

INFLECTION

PART XXXV — TURNAROUND MODEL

PART XXXVI — THE DATA HIERARCHY

PART XXXVII — BUILD A FACT SHEET BEFORE FORMING AN OPINION

PART XXXVIII — THE MULTI-AI RESEARCH ARCHITECTURE

PART XXXIX — AGENT RESPONSIBILITIES

Source Agent

Business Analyst

Financial Forensics Agent

Valuation Agent

Quant Agent

Governance Agent

Red-Team Agent

Synthesis Agent

PART XL — WHY THE AI AGENTS SHOULD WORK INDEPENDENTLY

PART XLI — MASTER AI PROMPT

PART XLII — THE FINAL COMPANY RESEARCH CARD

PART XLIII — MONITOR THE THESIS, NOT THE SHARE PRICE

PART XLIV — BUILD A THESIS-INVALIDATION SYSTEM

PART XLV — BACKTESTING THE MODEL

PART XLVI — MULTIBAGGER RESEARCH IS NOT ABOUT PREDICTING PRICE

PART XLVII — THE MOST INTERESTING SIGNAL: MULTIPLE INFLECTION

PART XLVIII — WHAT NOT TO DO

PART XLIX — THE COMPLETE RESEARCH PIPELINE

FINAL PRINCIPLE