Friends,
Today’s webinar starts at 4 pm EST. Find the Zoom link at the end of this email.
This one is in the spirit of “build in public”. I’ve been using Claude to sandbox some automation around identifying and classifying potential option trades.
It’s a raw look at what I’m doing and how.
A non-exhaustive list of what I’ll cover
The “guitar pedal preset" framework where I’m teaching Claude to listen like me
Cross-sectional filtering to find the 10-20% of tickers worth trading
The power of parallel coordinates charts for inspecting data beyond 3 dimensions
Towards a “recommendations” table with tight/loose fits
The custom instructions loop I’m employing within the “project”
The cross-referencing tool I built in Excel
Discussion of pitfalls, concerns, and extensions
I’m operationalizing a process. The goal of the moontower.ai app is to give the same opinionated lens on option markets that I enjoyed as a market-maker and PM. The cross-sectional framework reduces to notable vol surface signatures. That’s where you can direct your attention. You see which option markets are offering something unusual and then bring your context or bias to bear on whether it’s a deal you should pursue.
Discretionary trading is always “messy” in some sense. If it weren’t, it would be called systematic. Still, it requires extensive process and measurement in the trade lifecycle. The webinar will discuss what we look for, but the process of how we look for it and how to do so automatically should inspire or provoke.
(Just a thought — I’ve really been enjoying the progress on this because I’m starting to see how it can turn into something we ship in the app, which automates the funneling process and does so in a way that is ultimately educational so the very act of prospecting for trades makes you better at it AND able to narrate your reasons better. Something harder than it looks in options because of its dimensions — but something I’m convinced is teachable with the help of design.)