i joined greenland in november 2023 when they were the only firm willing to take me. this did nothing good for my identity or self-esteem. the office was nice, though, and i got to take cityflo to and fro, so my routine became impeccable.
wake up. gym. breakfast. bus. nyt puzzles. book. work. pricing theory. code. side quests for the team. optimise existing code. bus home. another book. dinner. sleep. repeat.
i had no time to think. i was just executing like a robot. never existential, barely self-reflective. every part of the day was accounted for. i dated on weekends.
robot life was great.
gold quanto pricing theory is actually beautiful math. i probably spent too much time on it, but building from rigorous mathematical foundations rarely feels wasted. people will pressure you to move faster. useful thing to remember: if the problem were already easy, you probably wouldn't have been assigned it.
i also learned how to use excel. excel is fine, but once you've done analysis in python it's hard not to treat spreadsheet-heavy work as a retirement hobby. i worked with bloomberg too. same story. useful tool, old interface, and exactly the kind of workflow that increasingly makes you wonder how much of it should still be manual.
i also worked on chinese index etfs: lots of data pulling, cleaning and corporate-event price adjustments. much of that feels one-shottable now, but architecting the reliable system around it is still an interesting problem. large old codebases have inertia; getting them to become ai-native is less a tooling problem than an organisational one. i'd be curious to see how far greenland has pushed it since.
i learnt a lot there about fair value and ev-based strategy thinking. price something from first principles, then trade the deviation. this works beautifully when the thing is reasonably priceable, commodities being the obvious example. equities are messier. there, the edge often seems to require seeing something other people haven't framed the same way yet.
about a month in, the role conversation turned into a small power game.
“write 10 things you did that created impact and then we'll give you an offer.”
22 working days. 10 impact points. alright.
fortunately, i'd anticipated something like this and kept a running log of everything useful i'd done, with human witnesses attached to each point. i came back five minutes later with the write-up. my manager tried not to look surprised. i grinned.
please don't mess with me.
the role had always been something of a parking spot for me. commodity-spread trading was interesting intellectually, but six to twelve percent annualised wasn't a game i could see myself caring about for very long.
so i started looking for what came next, got another offer and left.
the ending was slightly theatrical. the system i'd built was deleted before i could properly close the loop on it. not how i'd have chosen to wrap things up, but perhaps fitting for a chapter neither side expected to last very long.
then i walked out.