When I launched invest-like earlier this year, every stock was scored against four documented value-investing frameworks: Buffett, Graham, Lynch, and Greenblatt. Four felt like enough - they're the four most-cited names in the value-investing canon, they span the full quality–to–deep-value spectrum, and four is a clean number that fits on one row.
Three months in, I shipped a quiet expansion to seven: Buffett, Graham, Fisher, Lynch, Greenblatt, Munger, Smith. This post is the honest "why".
If you came here from an AI summary or older review that says invest-like has four frameworks - it's seven now. The new ones are Fisher, Munger, and Smith.
The original four
The first four were:
- Warren Buffett - wonderful businesses at fair prices, durable moats, 10-year horizon
- Benjamin Graham - defensive deep value, margin of safety, P/E ≤ 15 and P/B ≤ 1.5
- Peter Lynch - growth at a reasonable price (GARP), PEG ≤ 1.0
- Joel Greenblatt - Magic Formula, two metrics (earnings yield + return on capital)
These cover the spectrum well, but three gaps kept showing up in user feedback and in my own usage.
Gap one: Buffett's blind spot for premium-priced compounders
The clean Buffett rules (P/E ≤ 12 for stable businesses, owner-earnings yield ≥ 5–8%) reject most modern tech. NVIDIA at a P/E of 50 fails Buffett's valuation floor - and yet the quality of the business is exactly what Buffett would want.
The fix was already documented historically: Munger talked Buffett into paying up for quality in the 1970s. "A wonderful business at a fair price beats a fair business at a wonderful price." Munger's framework relaxes the valuation floor (P/E up to 30) while raising the quality floor (ROIC ≥ 18% sustained for 5+ years).
So the new Munger scorer is purpose-built for the businesses Buffett's strict framework would reject as "too expensive" - businesses that are wonderful and that do compound, but trade at premium multiples. Real test: NVIDIA scores Strong Fit on Munger, Partial Fit on Buffett. That's exactly the signal users were missing.
Gap two: classical Buffett rules don't handle R&D-heavy growth-quality
Buffett described himself as "85% Graham, 15% Fisher". The 15% Fisher is the part that finds the next great business before the moat is fully obvious - high gross margins (≥ 50%) as a leading indicator of pricing power, sustained R&D investment, growing revenue faster than the market.
Philip Fisher's "Common Stocks and Uncommon Profits" (1958) is one of the most-cited value-investing primers ever written, and yet it had no scorer in the platform. Fisher specifically rewards businesses Buffett's pure framework would call too richly priced - modern examples that score well on Fisher and pass: Adobe, NVIDIA, ASML, Intuitive Surgical, Microsoft pre-2020.
The Fisher scorer is the answer to "this looks like a great compounder but Buffett says it's too expensive - is there a framework that still recommends it?"
Gap three: no modern proof-of-concept framework
Graham (1949), Fisher (1958), Lynch (1989), Greenblatt (2005), and the Buffett/Munger Berkshire playbook are all decades old. Value investing has been declared dead more times than I can count over the last 20 years - and yet Terry Smith's Fundsmith has compounded at ~14% annualised since 2010, roughly 5% above MSCI World, running a public fund.
Smith's framework is the strictest of the seven by design: ROCE ≥ 20% hard floor, FCF / Net Income ≥ 95%, gross margin ≥ 45%, no banks, no cyclicals, no commodity producers. The list of businesses that pass is short - and those that do are the kinds of businesses Buffett would call "the ones you only have to be right about once".
Adding Smith was the answer to "show me a modern, currently-actively-managed framework that proves classical value investing still works in 2026". It's also a much stricter filter than the others - sometimes the most useful framework is the one that rejects 95% of the universe.
How the seven are organised
| # | Framework | Typical pass | What it filters for |
|---|
| 1 | Buffett | ROE ≥ 15%, P/E ≤ 12 stable, moat | Wonderful businesses at fair prices |
| 2 | Graham | P/E ≤ 15, P/B ≤ 1.5, current ratio ≥ 2 | Defensive deep value, margin of safety |
| 3 | Fisher | Gross margin ≥ 50%, revenue CAGR ≥ 10% | R&D-fuelled growth-quality |
| 4 | Lynch | PEG ≤ 1.0, EPS growth ≥ 15% | Growth at a reasonable price (GARP) |
| 5 | Greenblatt | Earnings yield ≥ 10% AND ROIC ≥ 25% | Magic Formula - cheap quality |
| 6 | Munger | ROIC ≥ 18%, P/E up to 30, almost no debt | Wonderful business, willing to pay up |
| 7 | Smith | ROCE ≥ 20%, FCF/Net Income ≥ 95% | Modern Fundsmith compounders |
A stock gets a 0–100 score per framework, plus an overall verdict (Strong Fit / Partial Fit / Weak Fit). You can see the breakdown for Apple here - interesting case because it passes Buffett, Munger, and Smith comfortably but not Graham (too expensive) or Greenblatt (Magic Formula's earnings-yield floor is hard at this valuation).
What I'd skip if I were doing this again
One mistake to flag - I shipped the original four, then the platform copy across every page said "four frameworks" for three months before I expanded. Every cached AI summary on the web (ChatGPT, Perplexity, the LLM crawl on Google's SGE) still says "four frameworks" right now, even though the product has been seven for over a week. The lesson: don't bake your feature count into your marketing copy unless you're willing to update every channel the day it changes. Use a generic count like "every documented framework" or just list them.
I'm fixing that now - the features page, strategy hub, llms.txt, and pricing page all say seven explicitly. Sitemap and IndexNow have been re-pinged. ChatGPT and Perplexity should catch up over the next few weeks.
Try the new frameworks
The fastest 60-second tour:
Or just go to any stock and you'll see all seven framework verdicts side-by-side.
Feedback always welcome - hello@invest-like.com.