StockScreener

My name is Maksym Misichenko, I'm based in Kyiv, Ukraine. 10 years in marketing, Python and process automation as a hobby that got out of hand. You can learn more about me on the author page and about the project on the about page.

StockScreener started as a tool I built for myself. I needed two things: a way to systematically scan NASDAQ for swing trading entry points, and a way to actually understand why a stock moved — not just that it moved. Raw news wasn't enough. I wanted context, analysis, and multiple perspectives — so I went deeper.

The signal system is built entirely on math. Market regime, sector strength, fundamentals, technicals, scoring — every signal goes through a 7-step pipeline where each decision is based on formulas and backtested data, not opinions. At the very end, four AI models review the signal — but they don't influence it. Their role is purely advisory: here's what the algorithms found, and here's what neural networks think about it. I built it this way because that's exactly what I used to do manually — find a setup, then go ask AI assistants what they thought.

The news analysis works differently. Every article is sent to four leading AI models — Claude, Gemini, ChatGPT, and Grok — with the same prompt, no exceptions. The original article is the source of truth. The models are explicitly forbidden from hallucinating: if they can't back a claim with facts, they can't use it as an argument. No model gets special treatment, no topic gets special framing. They analyze, they debate, they sometimes change their minds — and you see the full reasoning.

Today this project publishes 100+ original financial news articles from around the world, each wrapped in AI analysis and a debate between 4 leading LLMs. Every day. In 16 languages. It's free, and I think it's genuinely worth your time.