Hong Kong
A record of restaurants that are actually good — and a few that aren't. No sponsorships. No star chasing. Just the food.
Read the reviewsTKL
Hard to explain why this one keeps coming back. Something in the production — the space between things rather than the things themselves.
Wan Chai, Hong Kong
Omakase that punches above its price point. The chef does something interesting with aged fish — nothing gimmicky, just controlled. Worth the booking wait.
July 2026Central, Hong Kong
A classic that earns its reputation. The roast goose is as good as advertised. Go at lunch to avoid the tourist dinner crowd — the room is different.
June 2026Sheung Wan, Hong Kong
Northeastern Thai done right. The larb here has actual heat and proper fermented funk — not the sanitised version. Loud room but you don't notice after the first round.
May 2026Other interests
Session logs, hand history, and GTO notes from the tables in HK and Macau.
Read →Game logs, training notes, and whatever the Hong Kong pickup scene looks like that week.
Read →Strategy consulting. Thought leadership. A quant model nobody asked me to build.
Read →What I've read recently and whether it was worth it. Across history, finance, journalism, and whatever else.
Read →Things I'm actively studying — AI, markets, history. Notes on what's clicked and what hasn't.
Read →What's on the playlist, what I'm watching, what's been holding my attention. Updated monthly.
Read →Cocktail bars, dive bars, whatever's worth a second visit. Same rules as the food log.
Read →Cities and trips, city/country level. Not a full itinerary — just what stuck.
Read →Ryszard Kapuściński
Dispatches from post-colonial Africa. Kapuściński writes like a novelist but it's all real. The detail on how power actually moves through a place — not the official version — is the whole point.
July 2026Mike Duncan
The Republic's slow collapse before Caesar. What's interesting isn't who broke it — it's the decade when the norms stopped holding and everyone could see it but nobody acted.
June 2026Gregory Zuckerman
Simons and Renaissance. The actual story is less romantic than the myth — it's about data infrastructure and hiring mathematicians, not about a genius insight. More instructive that way.
May 2026The SCA gives CLP a permitted ~8% return on fixed assets. That sounds safe until you realise the reference rate that makes 8% attractive keeps shifting. EnergyAustralia is the more interesting drag — three years of write-downs on a business they overpaid for during a different energy regime. The question isn't whether management knows this. It's whether the structure allows them to do anything about it fast enough.
Continue reading →Revenue pools in AI infrastructure are migrating toward whoever controls the transaction layer, not the model layer. x402 is early but the direction is right: if agents are making payments autonomously, the payment rail is infrastructure. That's a different business than anyone was building for two years ago.
Continue reading →More deals, lower average premiums. The obvious read is more competition for assets. The less obvious read is that buyers are getting more disciplined after the 2021–22 vintage proved expensive. The cross-grid convergence thesis holds but the arbitrage is getting priced in faster than it was.
Continue reading →"The sun also rises."
"Africa is a vast territory, and whatever you say about it the opposite is also true."
"The Roman Republic did not die because it had to. It died because enough Romans chose short-term advantage over the long-term health of the system they depended on."
"Yeah, I will get to the top."
This is a personal record of restaurants I've been to and found worth writing about. The ratings are my own and calibrated against each other. Nothing is sponsored or gifted.
When a place is bad I'll say so. When it's good I'll say why specifically, not just that it's good.