Turning Investing Books Into Python Screens

By summer 2024 I had read enough investing material that leaving everything as notes started feeling silly. Greenblatt had a clean formula. Lynch had the practical curiosity. Buffett and Munger had the owner-earnings and stupidity-avoidance pressure. Pabrai kept talking about asymmetry. Damodaran made assumptions impossible to hide from.

So I did what I usually do when something starts occupying too much mental space: I tried to make it executable. Python, yfinance, pandas, tickers, regions, ranks, tables, weird missing fields, and the usual battle between a nice idea and the data available in the real world.

The point was not to worship a formula. That would be too easy and probably expensive. A formula is a net. It catches some things, misses others, and occasionally pulls up something that smells suspicious. I wanted a first-pass machine that could remove obvious noise and give my attention a better battlefield.

Lynch mattered because he kept the work close to reality: products, customers, cycles, boring businesses, crowded stores, unloved categories, simple observations. Greenblatt gave me a ranking mechanism. Munger kept standing behind the whole thing with the emotional warmth of a tax audit, asking where the stupidity was hiding.

That mix suited me. I like abstraction, but I do not trust abstraction when it floats away from the object. A company is still a business. A screen can point. It cannot understand for you. The useful part was letting Python do the repetitive work while keeping the judgment layer alive and slightly suspicious.

Looking back, this was when the reading started becoming infrastructure. I was not only collecting views from great investors anymore. I was trying to make their questions collide inside a system I could run, inspect, distrust, and improve.