Factor modeling, structured data pipelines, and probability scoring — applied to the most publicly available, outcome-measurable dataset in the world. Built in public. Open to anyone curious enough to look under the hood.
Most retail investors make decisions based on headlines, hot takes, and gut feeling. Institutional traders don't. They use factor models — systematic frameworks that score assets across multiple independent signals and combine them into a single probability estimate.
That gap has always existed. Data science can close it. The stock market is the perfect laboratory: the data is public, the outcomes are measurable, and the feedback loop is fast. Unlike most academic datasets, people actually care about the results.
Every score is generated from structured, rules-based calculations using real market data. No gut feeling. No narrative. No AI opinion dressed up as analysis. Each factor is independently sourced, weighted, and combined.
Every data source is named, documented, and independently auditable. Everything here can be replicated.
Google Gemini generates the directional call (LONG or SHORT), price target, time horizon, and the one-sentence rationale on the result screen. That's it.
Every weighted score in the model is derived from structured, rules-based calculations using real market data. Gemini does not control any of the five factor scores. It reads the data we've already processed and produces a human-readable output.
AI outputs can be wrong, inconsistent, or confidently mistaken. We know this. That's why the scores are anchored in data — not in what an AI thinks about a stock.
Every prediction made on this site is logged with a timestamp and entry price. In 90 days, we score it — WIN, LOSS, or PUSH. Over time, that dataset becomes a public accuracy record.
The model will be wrong. The question is how often, and whether systematic analysis beats random chance. That's the experiment. You're watching it run.
Anyone curious about how data science actually works in practice. Retail investors who want a structured alternative to noise. Aspiring analysts who want to see a live factor model. Students who learn better by building than by reading.
If you're interested in the methodology, the data, or the craft behind this — follow along. The model is an open experiment, not a closed product.
A data scientist with an MS in Data Science, specialized in business analytics. The Odds Algo is a continuation of that craft — applying factor modeling and structured data analysis to a domain that most people navigate on instinct.
The entire stack is built with the assistance of Claude (Anthropic) — for architecture decisions, code review, debugging, and feature development. Every meaningful decision is made by a human. Claude is the engineering partner that helps execute faster and more rigorously.