In one sentence: Phil teaches the sticker-price calculation by hand: grow EPS 10 years, apply a future P/E, discount at 15% a year (divide by 4), then halve it for the margin of safety, after first taking apart the idea that risk is volatility.
Key ideas
- Mastery stages. Unconsciously incompetent, consciously incompetent, consciously competent, unconsciously competent. Buffett supposedly judges a business in about 15 minutes. Practice widens the "canyon". [00:00–08:00]
- Diversification is a hedge against ignorance. Phil paraphrases Buffett: you diversify because you don't know what you're doing. Robo-advisors and advisors sell MPT portfolios. [07:00–11:00]
- Beta is not risk. MPT measures risk by how much a price moves against the S&P 500. Buffett's retort: buying Gildan at $15 is less risky than at $45 if the business hasn't changed, yet beta rose after the fall. Risk comes from not knowing the business, as with a 12-year-old driving you to work. [10:30–17:00]
- Price isn't value. The Maserati dealer markup and the mink-coat buyer are the examples. Shiller's Irrational Exuberance is the counter to Malkiel's Random Walk. [19:30–25:00]
- Lucky monkeys in one zoo. Buffett's Graham-and-Doddsville talk argues that great investors who share a method aren't random. [25:30–27:30]
- The five inputs. (1) TTM EPS (trailing twelve months earnings per share, e.g. from Yahoo Finance), (2) growth rate (your estimate or analysts' five-year number, whichever is lower), (3) future P/E, (4) a minimum acceptable rate of return (MARR) of 15%, (5) years: 10. [31:00–47:00]
- Future P/E = the lower of twice the growth rate or the historical high P/E. Reasoning: the long-run average P/E of about 15 and growth of about 7.5% give the "2× growth" rule of thumb. You sell only in a good market, so use an optimistic multiple, then compensate with the discount. [35:30–41:00]
- Why a multiple at all? Earnings are real money to the owner. Phil's class exercise has buyers and sellers haggle, and it keeps landing within a turn or two of 15× earnings. [41:30–44:30]
- MARR of 15%. It's well above the roughly 2.1% 10-year Treasury, and it's the same for every business. Phil says he doesn't adjust for risk. [44:30–47:30]
- The arithmetic (Apple, 2015). EPS $8.65 grown at 14% for 10 years is about $32.07. × P/E 28 ≈ $897 in 10 years. ÷ 4 ≈ $224 (the sticker). × 50% ≈ $112 (margin-of-safety price). Dividing by 4 works because 15% doubles money about every five years. [47:00–53:00]
- Dated and not advice. Phil says to do homework first and that the Apple numbers will be wrong by the time they are published. He also notes Icahn's $220 estimate agrees. [52:00–54:00]
How it maps to RuleOne
- This is the logic in
ruleone/valuation.py. Compare it line by line with the five inputs above to see where RuleOne deviates (for example the growth cap, or which P/E cap it uses). - The stock page shows sticker and margin-of-safety prices, and the screener sorts by distance below MoS.
- The agent should write down the growth and P/E it used so a human can change them.
Buffett, Munger and Graham links
- Buffett's "Superinvestors of Graham-and-Doddsville" (1984) is the talk Phil cites. Phil places it in 1988 on the audio, which looks wrong, so use the 1984 source.
- Buffett on future cash flow as the value of a business: his 1992 letter has the "discounted cash flows" passage. Check it before quoting.
- Taleb's The Black Swan is Phil's recommended book on fat-tailed markets.
Words to know
- TTM EPS: earnings per share over the trailing twelve months.
- P/E ratio: price divided by earnings per share.
- MARR: minimum acceptable rate of return, Phil's 15% a year.
- Sticker price: the estimated value today if growth and P/E come true. MoS price: half of it.
Try this
On a stock page (or any ticker you know), take today's TTM EPS and the analyst growth rate. Grow EPS 10 years, multiply by the lower of 2× growth or the historical high P/E, divide by 4, then halve. Compare your answer to the page's sticker price. If they differ, work out which input did it.
Check yourself
- Why divide the 10-year price by 4?
Answer
At 15% a year money doubles about every five years, so ten years is two doublings, meaning a factor of about 4. - Which P/E do you use for the future price?
Answer
The lower of twice the growth rate and the company's historical high P/E. - What's wrong with using beta as risk?
Answer
It measures price movement, not knowledge of the business. A good company that fell a lot looks "riskier" even though it's cheaper.
Short quotes
"Price is what you pay, value is what you get." (Phil, ~27:00, auto-transcribed)