Most financial analysis stops at ratios — revenue growth, margins, ROE. Those are useful, but they don’t tell you how reliable or persistent a company’s performance actually is. I built a structural feature layer on top of SEC financial data to focus on three things: earnings quality, stability, and consistency over time. Instead of just asking “what are the numbers today?”, the goal is to understand how those...
secfsn_plus is a canonical financial feature engineering layer built on top of SEC fundamental data. The project takes a wide SEC financial panel indexed by company and period, then transforms it into research-ready outputs: accounting ratios, factor scores, composite signals, screening outputs, and ML-ready feature matrices. The important design choice is separation. secfsn handles the base SEC data layer. secfsn_plus handles deterministic feature engineering. Future layers can handle...
I recently finished the first version of a philosophy-oriented NLP research pipeline focused on shareholder letters. The idea behind philo_nlp is simple: long-form corporate communication often reflects an underlying philosophy. In this first implementation, the reference philosophy is Berkshire Hathaway and Warren Buffett shareholder communication. The project is split into three repositories. Berkshire_Letters builds the Buffett reference corpus by downloading and normalizing Berkshire shareholder letters into a reusable...
I built a local research dashboard to make Commitments of Traders data easier to review across a focused futures market universe. The goal is simple: take raw CFTC COT data, clean it, match it to a configured set of markets, and surface where current positioning sits relative to recent history. This is not a trading signal tool. It does not predict prices, recommend trades, or replace market judgment....
Most macroeconomic analysis lives at one of two extremes: This project sits somewhere in the middle. I wanted a clear, repeatable, and modular way to explore macroeconomic regimes using publicly available FRED data — something that supports learning, experimentation, and practical monitoring without pretending to predict the economy. That’s what this repository is. Here’s a link to GitHub. What This Project Is This repo is a macro research...
As part of a larger NLP project focused on analyzing shareholder communications, I needed a clean, consistent corpus of historical letters to shareholders. One obvious and valuable dataset is the collection of Chairman’s Letters from Berkshire Hathaway, which spans nearly five decades. While the letters are publicly available on Berkshire’s website, assembling them into a single, NLP-ready dataset isn’t quite as simple as it sounds. Here’s a link...
Working with SEC fundamentals data is powerful—but rarely simple. Raw filings are large, fragmented across periods, and difficult to transform into something usable for research, screening, or visualization. To solve this, I built SEC FSN, a lightweight Python-based data engineering and analytics pipeline for working with SEC EDGAR Financial Statements & Notes (FSN) data. The goal was simple: create a transparent, reproducible fundamentals foundation that works equally well...
Background One of the first Python and finance books I ever read was Algorithmic Short Selling with Python.Hands down, it’s one of the best books I’ve ever read that blended Python with finance — teaching both the programming and financial concepts in a clear and practical way. I continue to come back to this book over time and recently noticed that some of the included Jupyter notebooks had...
Blackjack is one of the most widely studied games in both casinos and academia. The rules are simple, but the strategy is subtle, and the possibility of card counting has made it a fascinating testbed for probability, game theory, and decision-making. The question at the heart of this project: can a machine, using reinforcement learning (RL), teach itself how to play blackjack — and even rediscover professional strategies...
In Part 2, we trained a neural network to price options using the Cox-Ross-Rubinstein (CRR) model. It learned to predict the fair value of an option using inputs like strike, time to expiry, interest rate, and volatility. Now, we’ll use that trained model to do something powerful:Scan live option chains and identify mispriced vertical call spreads — a core strategy for directional or volatility-based options trading. This is...
All rights are reserved.
