Mohnish Pabrai Owns Four Stocks. Our Model Rates Two of Them Extremely Overvalued.

Pabrai is now on our superinvestor page, and he arrives with the fewest positions of any manager we track. Li Lu’s Himalaya Capital is next with 7; Warren Buffett runs 26. His book is also the smallest — about an eighth of the next-smallest 13F on the roster (Fairfax, at $2.6B). “Very few bets, large bets, infrequent bets” is not a slogan he uses on podcasts. It is what the filing says.

The last four filings read like someone quietly narrowing down rather than adding ideas. He owned three offshore drillers a year ago; now he owns one.

Here is where it gets useful. Every stock on Finrys gets the same treatment: an 8-point health check, a predictability score, and a fair value built from normalised cash flows. We ran his four holdings through it on today’s prices. Nobody tuned anything for this article.

Our engine classifies all three of his big holdings as cyclical commodity businesses, and it scores their predictability at 1, 1 and 2 stars out of 5. Predictability measures how steadily a company’s own history has compounded — the underlying number is a coefficient of variation, and lower is steadier.

The interesting holding is the tiny one. In the interview Pabrai walks through Kaspi.kz in detail — the “WeChat of Kazakhstan”, a failing bank rebuilt into a super app by a founder-CEO who owns roughly 40% of it, $2bn of annual cash flow out of a country of ten million people, and a moonshot in Turkey that has eight times the population. Then the filing shows he owns 1,702 shares.

That is a starter position, or a US-listed sliver of something he holds elsewhere. Either way it is the one name in his book our engine also likes — and it is the cheapest thing in this article.

This is the part a superinvestor page is actually for. At the same quarter end, five other managers we track held Kaspi — and the two new positions in the name were opened by Pabrai and by Prem Watsa’s Fairfax, at 0.35% of its portfolio. Baillie Gifford, the largest holder by a distance, added 12.8%. Dodge & Cox added 20.3%. ARK trimmed a third.

Pabrai’s rule is blunt: the things people discuss on YouTube and podcasts are exactly where he will not invest. Look at who is standing next to him in his three big names and you can see what that means in practice.

Asked about Adobe, Pabrai gives a clean answer: if you have strong conviction in the minimum cash flows over the next five to fifteen years, discount them back and the decision is obvious. If you don’t, move on. For him, Adobe is a pass — not because it is bad, but because he cannot get conviction on the AI question.

Our engine has no such doubt. Adobe passes 8 / 8 pillars, earns 29% on capital at a 89.4% gross margin, has shrunk its share count 13.1% in five years, trades on 15.9× earnings, and carries a fair value of $755.54 against a $285.50 price. Verdict: Strong Buy — the widest gap on this page.

Pabrai names one item as absolutely non-negotiable: the integrity and capability of management. His live example is NVR, the US homebuilder famous for buying back 80–90% of its stock over two decades. He walked away, because roughly 40–50% of those repurchased shares end up in management’s own pockets. “Why would you want to be slightly in bed with a crook?”

Now look at what our model sees. NVR: 15.0× earnings, 40.3% return on equity, zero debt, share count down 18.8% over five years, 6 / 8 pillars, health strong. Every quantitative box is ticked. The only mark against it in our data is price — fair value $5,008.57 against a $6,361.29 quote — and a profit line that has grown just 2% a year.

On when to sell a great business, Pabrai is precise in a way most people are not. Costco at roughly 50× trailing earnings has, in his view, never been egregiously overpriced. His number for egregious: 250× trailing normalised earnings. Anything below that, for a business with a genuine moat, he holds.

Our engine, pointed at the same names, is far less forgiving — because it is answering a buyer’s question, not an owner’s.

His concrete suggestion for someone who would normally dollar-cost-average into an index: don’t buy the S&P here, buy BRK.B and treat it as your index. Roughly 40% of the market cap is cash, another quarter or so is good listed businesses, and if a dislocation comes, someone at Berkshire gets to spend that cash while your index fund can only fall with the market.

Our model values Berkshire on book value — the same yardstick Buffett himself uses for buybacks — and reads 1.26× book, 10.5× earnings, a fair value of $551.18 against a $502.59 price. Verdict: Buy. It is one of the few large caps in this entire article that our engine does not consider expensive.

And the $30bn Google position Buffett confirmed was his own call? GOOGL scores 7 / 8 pillars, a 24.8% five-year ROIC, a 45.6% net margin, 16.9× earnings — and a fair value of $636.27 against a $340.42 price. Strong Buy. Pabrai’s caveat still stands, though, and it is a good one: Buffett makes that bet because his universe is tiny. At $100bn under management there would be no Google position at all.

Pabrai concedes the memory business looks insulated — three players, enormous barriers, and a Micron CFO telling him that even with all the patents, engineers and process people, they aren’t sure they could replicate one of their own fabs, “because there’s a part of this business that’s black magic”. He still passes. Where does it go three years out? Nobody knows. Too hard pile.

Our numbers say something similar in a colder voice. Micron’s five-year ROIC is 7.0% despite a 55.9% trailing net margin. Free cash flow per share is $0.62 — the stock is $931.27, which is roughly 1,501× free cash flow. Five-year FCF growth is −41.3% a year. Predictability: 1 star, and a composite CV of 3.49 — the noisiest history on this page by some margin. The boom is in the price. It is not yet in the cash.

You are not going to run a four-stock portfolio, and you shouldn’t. But four habits from this interview translate directly into how our tools are meant to be used.

Pabrai’s filing is not a shopping list. Two of his three large positions are, on our numbers, expensive businesses with deteriorating trailing results and one-star predictability; copying them without his cycle thesis would be copying the position and leaving the reasoning behind. The one holding our engine and his thesis agree on is the one worth $107.68 a share and, in his book, $147,461 — which tells you how early he thinks he is, or how much of it sits outside a 13F.

The transferable part is the discipline. He passed on Adobe, which our model rates the single best risk-adjusted price on this page, purely because he could not get conviction on the cash flows. He passed on SpaceX, on Micron, on the entire AI trade, on the passive-bubble debate — “why ask the address of a home you are never going to visit?”. Then he put 99.96% of his US book into three things he thinks he understands. Our job is to make the arithmetic fast and honest, including where it disagrees with him. The conviction is still yours to supply.