Methodology
StockRanks is a monthly ranking system for U.S. listed stocks. Rankings are produced by machine learning models trained on company fundamentals, price, and volume. The process is systematic: rank the universe, publish, then track. Rankings are generated on the last Friday of each month and published the following Monday.
Universe
Each run covers approximately 3,500 liquid U.S.-listed common stocks, screened for practical investability.
Data and models
Each monthly rank is a machine-learning forecast of expected relative performance over the next 12 months. Models are trained on:
- Fundamentals — income statements, balance sheets, and cash flow statements
- Price — historical price-based features
- Volume — trading-volume features
The same data definitions are applied across the universe each month so the ranking is comparable from name to name. We do not publish model architecture or a full feature list.
Monthly process
The universe is ranked once a month for expected relative performance over the next 12 months. The machine learning models are run on the last Friday; results are posted the next Monday, before that month’s forward returns are known. Live prices used for the track record refresh after the close on weekdays (ET).
Published files are not rewritten. Performance uses the original names and buy prices.
Five rating categories
Each month, stocks are assigned to one of five tiers:
- Strong Buy (Top 2%) — Expected to strongly outperform the market.
- Buy (Next 8%) — Expected to outperform the market.
- Hold (Middle 70%) — Expected to perform in line with the market.
- Sell (Next 15%) — Expected to underperform the market.
- Strong Sell (Bottom 5%) — Expected to significantly underperform the market.
Strong Buy and Strong Sell are uncommon by design — they are the tails of the monthly ranking.
Equal-weight tracking
The Strong Buy list is tracked as an equal-weight list versus the S&P 500 over a 12-month window. This is a measurement convention, not a claim about optimized portfolio construction.
Long-term, systematic approach
Rankings are rules-based and intended for a 12-month horizon. They are not day-trading signals and they are not a substitute for your own research.
Background
StockRanks was built by a data scientist who spent years on fraud-detection systems at U.S. fintech firms. The same emphasis on patterns, anomalies, and repeatable rules is applied here with machine learning on fundamentals, price, and volume — with the aim of ranking companies more consistently than a narrative-driven process.
How this record is kept
- Rankings are published before that month’s forward returns exist.
- Historical ranking files are not revised after publication.
- Each month’s Strong Buy list remains on the performance page.
- Performance uses the original published names and buy prices, equal-weight, versus the S&P 500.
FAQ
When are rankings updated?
Rankings are generated on the last Friday of each month and published the following Monday.
What is the investment window?
Each month’s ranks are evaluated over 12 months versus the S&P 500.
How is the Strong Buy list measured?
Equal-weight. It is not an optimized or leveraged portfolio.
Why compare to the S&P 500?
The S&P 500 is a simple, investable U.S. large-cap benchmark for the same holding window.
Do historical rankings change?
No. Files are not revised after publication. Performance uses the original published names and buy prices.
Is this investment advice?
No. Output is informational. Past performance does not guarantee future results.
How are ranks generated?
Machine learning models trained on fundamentals (income statements, balance sheets, cash flow), price history, and volume. Each stock is ranked for expected 12-month relative performance.
What is the universe?
About 3,500 liquid U.S.-listed common stocks, screened for practical investability.