Build a consistent market snapshot
The pipeline collects the price, volume, liquidity, and outcome data required by each supported model across eligible NYSE and NASDAQ equities.
Method / 01
From market close to a ranked research queue.
PatternRank is a five-stage research loop: establish the eligible universe, score the supported patterns, publish the queue, preserve the context, and measure what followed.
InputConsistent market + liquidity data
OutputA finite model-ranked queue
AccountabilityCompleted outcomes remain visible
The five-stage loop
A process you can review, repeat, and challenge.
The method is deliberately legible. Each stage narrows the problem and preserves enough context for the output to be evaluated later.
Apply the universe rules
Coverage filters remove securities that do not match a model’s structural requirements. ETB models add borrowability and liquidity constraints before ranking begins.
Score patterns inside the model
Each model evaluates the historical setup it was designed for. The result is a comparable score within that model—not a universal claim about the company.
Publish a finite ranked queue
After market close, the highest-ranked eligible setups appear with their confidence, model, horizon, date, and supporting context for independent review.
Close the loop with outcomes
When the review horizon completes, the measured result is attached to the original ranking and included in aggregate full-sample diagnostics.
Read the fields correctly
Confidence is context. Outcome is evidence.
Two numbers appear often in PatternRank. They answer different questions and should never be treated as interchangeable.
Model confidence
A relative score produced by the model at ranking time. It helps order setups inside the supported research universe. It is not a guaranteed probability of profit or a substitute for independent analysis.
7-day outcome
The measured price change seven calendar days after the ranking date, when data is available. It closes a consistent review window; it does not model a specific entry, exit, position size, or live portfolio.
Active research lenses
One product, distinct model questions.
Model labels stay visible because each universe and horizon has a different job. Combining them into a single black-box score would erase useful research context.
Short-horizon momentum
Ranks recent momentum behavior across the supported US equity universe and measures follow-through over a seven-day review window.
Liquidity-filtered quality
Adds easy-to-borrow and liquidity constraints for researchers who prefer a structurally tighter universe before evaluating momentum patterns.
Method boundaries
The ranking stops where your judgment begins.
PatternRank orders evidence. It does not decide whether a security fits your objectives, risk tolerance, time horizon, or portfolio.
Useful for
- ✓Building a consistent daily review queue
- ✓Comparing setups inside a defined model
- ✓Studying completed outcomes and regimes
- ✓Challenging a model with public evidence
Not designed for
- ×Personalized investment advice
- ×Intraday alerts or automatic execution
- ×Position sizing or portfolio construction
- ×Predictions with guaranteed outcomes
See the method under pressure
Start where the outcomes are already known.
The public track record is the fastest way to understand both the method and its limitations. Research use only. No investment advice, recommendations, or guarantees.