01 ‐ Project Overview - IMGitH/dvmax GitHub Wiki

Purpose

Build a machine learning engine that selects a basket of non-U.S. dividend-paying stocks with:

• High net dividend yield
• Low risk of dividend cuts
• Stable financial and price profiles
• Minimal U.S. estate tax exposure

Goals

• Publish a working E2E ML solution (learning + showcase)
• Generate meaningful portfolio profits (dividend income + total return)
• Lay the groundwork for a monetizable product

Core Principles

• Predict probability of dividend cut within 12 months (classification)
• Invest only in stocks with low cut risk + attractive adjusted yield
• Focus on non-U.S. stocks, via direct holdings or UCITS ETFs
• Emphasize stability, explainability, and practical trading constraints

Approach Summary

• Use global stock fundamentals + market performance + macro + sector
• Prioritize dividend yield + safety + valuation sanity
• Periodic rebalancing, minimal turnover