Stock research guides · Guide 7 of 10
Deep value and quality compounders: two ways to judge the same stock
Why a durable business can still look expensive: comparing a deep value framework with a quality compounder framework on the same company.
Key takeaways
- Deep value separates business resilience from the price you pay.
- A quality lens asks whether attractive returns can be sustained, and still weighs the price.
- A negative margin of safety means a model considers the price too high, not that a fall is certain.
- Compare the methods and assumptions behind two reviews, not their score totals.
Two careful investors can study the same company and reach opposite conclusions because they ask different questions. This guide explains how the deep value and quality compounder traditions weigh a business against its price, and how to read AI-generated reviews in each style without treating a label or a score as a verdict.
What an AI analyst review is, and is not
On dotQuant, AI analysts are AI-generated research perspectives, each with a fixed investing style. They are not human analysts: a review applies one style's criteria to the financial data and share price available when it was evaluated, and reports a label, a score and written reasoning.
A label is not a recommendation to buy, sell or hold, and a confidence percentage is not a probability of profit: 70% confidence does not mean a 70% chance of making money.
Check the evaluation date before comparing a review with anything else: prices move and new results arrive, so two panels can describe different moments.
The deep value tradition
Deep value investing traces back to Benjamin Graham, whose books Security Analysis (1934, with David Dodd) and The Intelligent Investor (1949) argued for buying only at prices well below a conservative estimate of value. That gap is the margin of safety: room for error or bad luck before a price paid becomes a permanent loss.
His tests for defensive investors favoured evidence over forecasts. They included positive earnings in each of the past ten years, current assets at least twice current liabilities for industrial companies, and a price of no more than 15 times average earnings and 1.5 times book value, or multiples whose product stays within 22.5: the rule behind the Graham number.
Graham number = √(22.5 × earnings per share × book value per share)Take a hypothetical company with earnings per share of 2 and book value per share of 20: its Graham number is √900 = 30. A share price of 24 sits 20% below that estimate; a price of 45 is 1.5 times it, with no margin of safety by this standard.
The estimate is conservative by design. It gives no credit for growth, brands or high returns on capital, so asset-light businesses with small book values tend to trade far above it, and it is meaningless when earnings or book value are negative.
On dotQuant, the AI analyst "Margaret Holloway" has the "Deep Value" style and the tagline "Buy a dollar for fifty cents." Its review is scored on earnings stability, financial strength and valuation, which compares the price with a conservative Graham estimate. The tests can disagree, so read each one: which periods support the earnings record, and whether current assets cover near-term obligations.
The quality compounder
A quality compounder earns high returns on the capital it employs, sustains them for years and can reinvest profits at similar rates. Durability is what compounds: a hypothetical business earning 20% on equity and reinvesting everything at that rate grows its equity from 100 to about 249 in five years, while one earning 8% reaches about 147.
A quality lens therefore looks for returns on equity and invested capital that stay high, strong operating margins, low debt and consistent earnings. Persistently high returns often point to a competitive advantage, the subject of the moats guide. Price still matters: the lens compares the share price with an estimate of intrinsic value, typically from a discounted cash flow model.
On dotQuant, "Howard Sterling" has the "Quality Compounder" style and the tagline "Wonderful businesses at fair prices, held for the long run." Its review is scored on fundamentals, earnings consistency and valuation (intrinsic value).
Reading the quality evidence carefully
A quality business can still look expensive. A price of 30 times earnings already builds in years of high returns and growth, so ask not only whether the company is good but what expectations its price includes, and what happens if growth merely slows.
Return on equity = net income ÷ shareholders' equityLeverage can flatter that ratio. Take two hypothetical companies with the same business and 1,000 million of assets: one funded entirely by equity earns 100 million, a 10% return on equity. The other borrows half and, after 20 million of interest net of tax, earns 80 million on 500 million of equity, a 16% return from a business that is no better, only more indebted.
Low debt does not guarantee liquidity either. A company with no borrowings can still owe suppliers, staff and tax authorities more than its cash and receivables cover, as the financial health guide explains.
A review that discusses several periods may quote cumulative growth, which is not an annual rate.
Cumulative growth = (latest ÷ earliest) − 1Annual growth rate = (latest ÷ earliest) to the power (1 ÷ years) − 1If a hypothetical company's earnings rise from 100 to 200, cumulative growth is 100%. Over five years that is about 14.9% a year, but five annual reports span only four years of change, which gives about 18.9%. Compare like-for-like periods before setting one panel against another.
When the two lenses disagree
Take a hypothetical company with a share price of 150, earnings per share of 5, book value per share of 20, a 25% return on equity, a 30% operating margin, debt to equity of 0.3 and a current ratio of 1.2. Earnings have risen every year for five years. Its Graham number is √(22.5 × 5 × 20), about 47, while a discounted cash flow model that assumes growth continues puts intrinsic value at 180.
| What is examined | Deep value lens | Quality compounder lens |
|---|---|---|
| Earnings record | Positive every year: passes | Rising every year: passes |
| Balance sheet | Current ratio 1.2, short of two to one | Debt to equity 0.3: low debt |
| Returns and margins | Not part of the test | 25% and 30%: strong |
| Value estimate | Graham number, about 47 | Modelled intrinsic value, 180 |
| Price of 150 | About three times the estimate | About 17% below the estimate |
| Conclusion | Stable earnings, no margin of safety | Strong business, modest margin of safety |
Neither lens is wrong; they value different things. The Graham estimate ignores growth and high returns, so an asset-light compounder rarely passes it; the quality estimate is only as sound as its growth assumption.
Negative margins of safety and score totals
Margin of safety = (estimated intrinsic value − share price) ÷ estimated intrinsic valueDefinitions vary: this common version divides by the estimate, some tools by the share price. Either way, a negative figure means the model's estimate is below the current price, so the framework considers the price too high under its assumptions. It is not a forecast that the price will fall: the estimate can be wrong, prices can fall as well as rise whatever a model says, and model estimates do not guarantee future results.
Assumptions drive the sign. If the model above assumes slower growth and its estimate drops from 180 to 125, the margin of safety moves from about 17% to (125 − 150) ÷ 125 = −20%: same company, same price, opposite conclusion.
Scores need the same care. Each framework uses its own criteria and maximum, so a hypothetical 14/20 from one analyst and 7/10 from another are both 70% yet measure different things; neither raw totals nor percentages can rank them. Agreement is not independent confirmation either: both reviews read the same company accounts and a similar price, so an unusual year or a data error flows into both.
Putting it into practice on dotQuant
The "Fundamentals Matrix" panel on each symbol page lists six AI analysts, each showing its name, style, a "BULLISH", "NEUTRAL" or "BEARISH" label and a score such as "14/20", with the evaluation date and time. Opening one shows its confidence percentage, written reasoning and score breakdown. To read the two styles in this guide:
- Check the evaluation date and time.
- Read the score breakdown before the label.
- Separate business from price: stability and strength versus valuation for "Margaret Holloway"; fundamentals and consistency versus intrinsic value for "Howard Sterling".
- Compare with the "Price to Graham number" and "Discounted-cash-flow margin of safety" rows in the Key Metrics "Valuation" group, allowing for different dates, prices and reporting periods. That panel's margin of safety divides the gap by market value rather than by the estimate, so a positive reading there is larger than the version above.
The other four styles appear in the guides on disruptive growth and activist quality and moats and macro momentum.
Without an account you can read the reviews and the Key Metrics panel on any symbol page, with end-of-day prices; live intraday prices need an account, the dotQuant desktop app and your own Interactive Brokers market-data subscriptions. See the AI analysts documentation.
Common questions
What is the difference between deep value and quality investing?
Deep value seeks a price well below a conservative estimate of a business's current worth, backed by stable earnings and a strong balance sheet. Quality investing seeks businesses that can sustain high returns on capital and may pay more for that durability, while still comparing price with intrinsic value.
What is the Graham number?
It is the highest price at which price to earnings times price to book stays within Graham's rule-of-thumb limit of 22.5. It ignores growth and is meaningless when earnings or book value are negative.
Is a bullish AI analyst label a buy recommendation?
No. It summarises how one fixed style reads the evidence on its evaluation date, without knowing your circumstances, and its confidence is not a probability of profit.
Further reading
- Stock valuation ratios explained: P/E, price to book and the DCF margin of safety.
- Profitability and growth metrics: returns on capital and annual versus multi-year growth.
- Disruptive growth and activist quality: two more styles and their DCF assumptions.
- SEC, Beginners' Guide to Financial Statements: where earnings per share and book value come from.
- FCA InvestSmart: the UK regulator's guidance on investment risks.