Benjamin Graham Deep Value Investing: AI-Powered Analysis Without Excel Hell
Benjamin Graham's deep value strategy is the foundation of modern value investing—buy stocks trading below intrinsic value with a margin of safety. Warren Buffett built an empire on these principles. But Graham's quantitative screens require endless financial statement analysis. Here's how AI turns 40 hours of spreadsheet work into 40 seconds of conversation.
Andrew Grosser
February 16, 2026 • 14 min read
January 2024: You're screening 2,847 publicly traded stocks looking for Graham-style deep value opportunities. The criteria are specific: trading below book value, positive earnings for five consecutive years, P/E ratio under 10, debt-to-equity below 1.0, current ratio above 2.0, and—most importantly—price below intrinsic value with a 30% margin of safety. You find 14 candidates. Now comes the hard part: calculating intrinsic value for each one using Graham's formulas, analyzing balance sheets, validating earnings quality, and comparing current price to your calculated fair value.
Benjamin Graham didn't just invent value investing—he quantified it. His 1934 book Security Analysis introduced systematic methods for identifying undervalued stocks using financial ratios and conservative valuation formulas. The Graham Number calculates a stock's maximum fair value based on earnings per share and book value per share. If current price is significantly below the Graham Number, you've found a deep value opportunity. But running these calculations across thousands of stocks in Excel? That's weeks of manual financial statement scraping, formula building, and error-prone data entry sign up free.
Or you use Sourcetable. Try it free.
Why Graham's Deep Value Strategy is Brutal to Implement in Excel
Graham's methodology isn't theoretical—it's deeply quantitative, requiring precise calculations across multiple financial metrics. The core principle is margin of safety: only buy when market price is substantially below intrinsic value, creating a cushion against estimation errors or business deterioration. Graham's famous formula for intrinsic value is: √(22.5 × EPS × Book Value per Share). This is the Graham Number—the theoretical maximum price a defensive investor should pay.
Let's say you're analyzing a beaten-down industrial stock—call it ABC Corp. Current price: $42. You need to determine if it's truly undervalued using Graham's criteria:
- Earnings Per Share (EPS): $4.80 for the trailing twelve months
- Book Value Per Share: $38.50 from the latest balance sheet
- Graham Number: √(22.5 × 4.80 × 38.50) = √4,158 = $64.48
- Margin of Safety: Current price $42 vs. intrinsic value $64.48 = 35% discount
ABC Corp trades 35% below Graham's calculated intrinsic value—a classic deep value opportunity. But that's just the first filter. Graham demanded additional safety checks:
- Earnings Stability: Positive earnings in each of the past 5 years (requires historical income statement analysis)
- Dividend Record: Uninterrupted dividends for at least 20 years (requires dividend history lookup)
- P/E Ratio: Must be below 10 (currently 8.75—passes)
- Price-to-Book Ratio: Below 1.5, ideally below 1.0 (currently 1.09—passes)
- Debt-to-Equity: Long-term debt should not exceed working capital (requires balance sheet ratios)
- Current Ratio: Current assets at least 2× current liabilities (requires liquidity analysis)
That's seven separate calculations requiring data from three different financial statements, historical records going back two decades, and manual verification of earnings quality. For one stock. Multiply by 2,847 and you have a full-time research job—which is exactly what Graham did, manually, without computers.
How Sourcetable Turns Graham's Framework Into Natural Language
Sourcetable doesn't replace Graham's methodology—it automates the grunt work. Import financial statement data (from your broker, financial APIs, or manual CSV files), and the AI handles every calculation, cross-reference, and historical lookup. You interact with Graham's framework the same way Graham himself would if he had an AI analyst on his team.
Instant Graham Number Calculation Across Your Watchlist
In Excel, calculating the Graham Number for 50 stocks means 50 manual square root formulas, each requiring you to pull EPS and book value from separate data sources. You'd create columns for ticker, price, EPS, BVPS, then write: =SQRT(22.5*C2*D2). Copy down, hope your data is accurate, manually compare to current price.
In Sourcetable, upload your watchlist with price and financial data, then ask: "Calculate Graham Numbers for all stocks." The AI instantly returns a new column with intrinsic values for every ticker. Follow up: "Which stocks trade below their Graham Number?" It filters down to 14 candidates. Ask: "Which have a margin of safety over 30%?" → 8 stocks flagged. Three questions, three seconds, and you've gone from 2,847 stocks to 8 deep value candidates without touching a formula.
Automated Quality Screens Using Graham's Defensive Investor Criteria
Graham distinguished between defensive investors (seeking safety and passive income) and enterprising investors (willing to do deeper research). For defensive investors, he prescribed strict quantitative filters. In Excel, implementing these requires nested IF statements, VLOOKUP formulas across multiple sheets, and constant manual updates.
With Sourcetable, simply describe Graham's criteria: "Show me stocks with positive earnings every year for 5 years, P/E under 10, price-to-book under 1.5, and current ratio above 2.0." The AI parses historical income statements, calculates ratios, and returns qualifying stocks in seconds. Change "P/E under 10" to "P/E under 15"—it recalculates instantly. That kind of dynamic filtering would require rebuilding your entire Excel workbook.
Real-Time Financial Statement Analysis
Graham emphasized reading balance sheets and income statements carefully. He wanted to see working capital exceeding long-term debt, consistent gross margins, and stable operating earnings. Doing this manually means downloading 10-Ks, copying numbers into Excel, building custom ratios, and tracking changes quarter over quarter.
Upload ABC Corp's financial statements to Sourcetable and ask conversational questions: "Does working capital exceed long-term debt?" → "Yes, working capital is $245M vs. long-term debt of $180M—passes Graham's safety test." "Has gross margin been stable for 5 years?" → "Yes, ranged from 38.2% to 40.1%, averaging 39.3%—stable." "Any red flags in the balance sheet?" → "Accounts receivable grew 24% last quarter while revenue grew 8%—worth investigating."
That last one—the AI flagging a mismatch between receivables and revenue growth—is something Graham would spot instantly but Excel won't tell you unless you specifically build a formula to detect it. Sourcetable's AI applies Graham-style skepticism automatically.
Tracking Margin of Safety Over Time
Graham's margin of safety isn't static—it changes as earnings evolve and market price fluctuates. A stock with a 40% margin of safety today might have only 10% next quarter if earnings decline or price rallies. Tracking this in Excel requires constant recalculation and manual chart updates.
Ask Sourcetable: "Track ABC Corp's margin of safety weekly for the past year." It generates a historical chart showing how the gap between market price and Graham Number has changed over 52 weeks. You instantly see that the margin of safety expanded from 15% to 35% over the past three months—not because intrinsic value increased, but because the market sold off while fundamentals stayed steady. That's exactly the kind of opportunity Graham hunted.
Screening the Entire Market for Graham-Style Deep Value
Graham's genius was systematic screening—applying rigid quantitative filters to eliminate speculation and focus on demonstrable value. Modern investors have access to thousands more securities than Graham did, making systematic screening even more powerful. But manual Excel screening is prohibitively slow.
Building a Graham Net-Net Screen
Graham's most extreme value play was the net-net strategy: buying stocks trading below net current asset value (current assets minus all liabilities). If a company's market cap is less than its liquidation value—literally worth more dead than alive—that's deep value with maximum safety. These opportunities are rare but explosive when found.
In Excel, calculating NCAV for thousands of stocks requires pulling balance sheets, summing current assets, subtracting total liabilities (not just current), dividing by shares outstanding, then comparing to market price. It's hours of data wrangling.
In Sourcetable: "Find all stocks trading below net current asset value." The AI calculates NCAV across your entire dataset, compares to market cap, and returns 4 candidates out of 2,847—true net-nets. Ask: "Which have been profitable for 3+ years?" → 2 remaining. You've just implemented Graham's most aggressive strategy in under 10 seconds.
Sector-Specific Graham Screens
Graham acknowledged that different industries require different benchmarks. Banks, for example, use different capital structures than manufacturers. Utilities have regulated returns. REITs focus on funds from operations rather than GAAP earnings. Applying one-size-fits-all screens misses industry context.
Sourcetable understands industry nuances through conversational queries: "Find banks with Tier 1 capital ratios above 12% trading below tangible book value." For REITs: "Show me REITs with FFO yields above 8% and payout ratios under 75%." For utilities: "Find utilities with dividend yields over 5% and rate base growth above 4%." Each query applies Graham's value principles within the appropriate industry framework—something that would require separate Excel templates for each sector.
Portfolio Construction: Building a Graham-Style Defensive Portfolio
Graham didn't just teach stock selection—he prescribed portfolio construction rules for defensive investors. Minimum diversification: 10 stocks. Maximum position size: 10% of portfolio. Mix of industries to avoid sector concentration. Regular rebalancing when valuations shift. Implementing this discipline manually is tedious and error-prone.
Automated Position Sizing and Diversification
Say you have $100,000 to invest and 8 Graham-qualified deep value stocks. Graham would recommend equal weighting—$12,500 per position—to maximize diversification. But if one stock has a 50% margin of safety and another has only 20%, should they get equal weight? Graham suggested adjusting for margin of safety while maintaining diversification.
Ask Sourcetable: "Allocate $100,000 across these 8 stocks, weighted by margin of safety, with a 15% position cap." It calculates optimal weights: ABC Corp (40% margin) gets $15,000 (capped at 15%), XYZ Inc (35% margin) gets $14,000, and so on down to the 20% margin stock at $10,000. Total: perfect diversification respecting Graham's safety principles.
Rebalancing Triggers Based on Valuation Changes
Graham advised selling when stocks reach intrinsic value or after a 50% gain, whichever comes first. Monitoring 10 positions for sell signals requires constant Graham Number recalculations and return tracking. Miss a signal and you're holding an overvalued stock that violates your discipline.
Sourcetable automates this: "Alert me when any position reaches its Graham Number or gains 50%." Two weeks later, ABC Corp rallies from $42 to $63—within $1.48 of its $64.48 Graham Number. The AI flags it: "ABC Corp is at 98% of intrinsic value. Consider selling to lock in 50% gain." You sell, redeploy capital into the next deep value candidate, and maintain discipline without manually tracking every position daily.
Adapting Graham's Principles to Modern Markets
Graham died in 1976. Markets have evolved—technology stocks with zero book value, intangible-heavy business models, global supply chains, crypto assets. Purist Grahamites argue his methods don't work anymore. Pragmatists adapt his principles: focus on undervaluation, demand safety margins, ignore market sentiment, buy when others panic.
Modified Graham Screens for Asset-Light Businesses
Graham's book value emphasis made sense in 1934 when businesses owned factories, inventory, and equipment. Today, the most valuable companies—software, platforms, brands—have minimal physical assets. Microsoft's book value is $31 per share; market price is $420. Traditional Graham screens would reject it as overvalued.
Modern value investors adapt by substituting owner earnings (Buffett's term) or free cash flow for book value in the Graham formula. Ask Sourcetable: "Calculate modified Graham Numbers using FCF per share instead of book value." It recalculates intrinsic values for all stocks using cash generation rather than balance sheet assets—Graham's logic applied to modern business models.
Combining Graham with Quality Factors
Pure Graham deep value often finds distressed companies—cigar butts with "one puff left," as Buffett put it. Modern quant strategies combine Graham's valuation discipline with quality factors: high return on equity, strong competitive moats, consistent free cash flow growth. This avoids value traps—cheap stocks that stay cheap because the business is deteriorating.
Sourcetable makes hybrid strategies trivial: "Find stocks below their Graham Number with ROE above 15% and positive FCF growth for 3 years." You get Graham's valuation discipline filtered through Buffett's quality lens—the best of both eras in a single query.
Key Takeaways
Benjamin Graham's deep value strategy is quantitative and systematic: buy stocks below intrinsic value with a margin of safety, using strict financial ratio filters to ensure quality and safety.
The Graham Number formula—√(22.5 × EPS × BVPS)—calculates maximum fair value. If current price is 30%+ below this, you have the margin of safety Graham demanded.
Traditional Excel implementation requires manual financial statement analysis, ratio calculations, historical data lookups, and constant updates—weeks of work for a full market screen.
Sourcetable turns Graham's framework into plain English: "Find stocks below their Graham Number" → 14 candidates. "Which have 5 years of positive earnings?" → 8 qualify.
Modern adaptations replace book value with free cash flow for asset-light businesses, and combine Graham's valuation discipline with quality factors to avoid value traps.
Frequently Asked Questions
If your question is not covered here, you can contact our team.
Contact UsWhat is the Graham Number and how do you calculate it?
What is Benjamin Graham's margin of safety principle?
What are Graham's defensive investor criteria?
What is a Graham net-net stock?
Does Benjamin Graham's value investing still work today?
When should you sell using Graham's strategy?
How many stocks should a Graham-style portfolio hold?
