The guy in the garage at 3 a.m.

Someone in a small apartment, somewhere, has spent four years reading everything published about drought patterns in the Horn of Africa. He does not work for a bank. He has no Bloomberg terminal, no compliance officer, no Christmas party with the other market makers. He just knows the thing cold.

On Kalshi, that man beats Wall Street.

Luana Lopes Lara, one of the exchange's founders, shared a number on the Cheeky Pint podcast that should embarrass anyone who worships the credential: less than 5% of matched liquidity comes from the big institutional market makers everyone assumes run the show. Over 95% comes from peer-style participants and tiny two-person shops. More than two thousand people quietly pricing weather, politics, and elections. The guys in the garage, her co-founder Tarek Mansour said, are the most crucial.

Consider what this breaks. The traditional exchange took two years to design a grain future, then called the same fifty firms it already knew. Slow. Clubby. Comfortable. Prediction markets need new liquidity daily, sometimes hourly, on questions no desk can spin up in an hour. The authority figure is useless here. The obsessive is priceless.

Depth in a narrow thing beats breadth in a prestigious one. The market doesn't ask for your résumé. It asks whether you were right, and it keeps score in public.

Most people spend their careers acquiring signals of competence. A smaller number acquire competence and let the score speak. The second group sleeps badly and gets rich. The first group has a nice title.

---

The billiard ball you'll never see

A man sits with his back to a billiard table. He cannot turn around. He must guess where a ball landed, and every guess is corrected by a single word: left, or right. That scene opens my latest essay, Confirmation Bias in Bayesian Dress: Authored Signals, Priced Priors, and the Weight of a Bet, and it is the cleanest picture I know of how a good mind should move.

One word at a time. The guess shifts. Nobody insists on being right at the start.

Contrast that with how most of us actually update. Neuroscientist Anne-Laure Le Cunff, who wrote Tiny Experiments, says that the best decision-makers write down both internal and external factors before they decide. Confidence out of ten. The metric on the page. She was describing a discipline against the oldest trap: mistaking the feeling of certainty for evidence of it.

Kahneman found the same thing in a supermarket. When a sign said "limit 12 per person," shoppers bought seven cans of soup. When the limit vanished, they bought three and a half. A number nobody chose, hanging off a shelf, anchored the whole decision. The billiard-ball player is corrected toward truth. The soup shopper is corrected toward a number some manager invented on a Tuesday.

Both are updating. Only one is updating on anything real.

The trick is not to stop having priors. You can't. The trick is to notice which word is moving your guess, and whether that word came from the felt or from the shelf.

---

Most of them will be worthless

Buy the book about the company everyone is buying. Sit with the contradiction.

The library shelf holds The Thinking Machine: Jensen Huang, Nvidia, and the World's Most Coveted Microchip, and the part worth carrying out of it is not the chips. It's Jensen's refusal to fight. Author Stephen Witt says that Nvidia's real moat was never speed. It was software nobody wanted to rewrite. Jensen had survived one knife fight early, a thirty-company brawl, and it scarred him so badly he spent the rest of his life avoiding crowded rooms. He built things that looked small and unprofitable and waited, alone, until they weren't. Two doctors used his first medical-imaging machine. Two.

Now hold that against something Edward Chancellor said on Howard Marks's podcast. The internet changed the world. Inarguable. And the vast majority of companies floated in 1999 to ride that change ended up worthless. Both things are true at once. The trend is real and the crowd chasing it is doomed. Companies added an "e" to their names and doubled. Today they add "AI" and do the same.

Two statements, Chancellor said, that are not inconsistent: AI will change the world, and most companies people are buying for AI will be worth nothing.

The tension nobody wants to hold is this. Being right about the technology tells you almost nothing about being right about the price. The naive investor collapses the two into one thrill. The internet is coming, therefore buy. The felt says nothing; he moved his guess on the shelf.

Jensen won because he stood in the empty room. The 1999 crowd lost because it ran into the full one. The world is loud right now about the same room. Ask which one you're standing in before you write the cheque.

---

My Ledger — notes on investing, books, and things I'm still figuring out. The Library is where I keep the books that shaped how I think. The Journal is where I work through ideas in public.

Mandelbrot — the private toolbox behind the thinking.

Hit reply if something here sparked a thought. I read every response and always write back.

---