Personal toy project
Quant Trading
Learning when a model should stay out.
A personal notebook of market-data preparation, dataset discipline and cautious strategy experiments.
Start readingIn this journey
The reading path
Follow the project from its first problem through the work that followed.
From no_trade Discipline to Reproducible Databento Ingestion
How a refusal-first research philosophy became a concrete data boundary with preserved Databento jobs, hash verification, bronze conversion, and quality reports.
From an MES/ES Silver Dataset to a Strict Training Matrix
How past-only contract selection and session filtering produced an MES/ES silver dataset, then a strict matrix with physically separated features, labels, metadata, and splits.
Why no_trade Baselines Made Positive Validation More Suspicious
How deterministic baselines, feature IC, and a strategy zoo selected no_trade, then train, concentration, and cost-stress gates rejected the first validation-positive candidates.
Decomposing the Train-Validation Gap with MBP Execution Reality
A forensic analysis of timing, contract, direction, fold, and cost fragility, followed by MBP-1 spread, latency, and adverse-movement calibration for OHLCV candidates.
Confirming no_trade with Research Gates and Multi-Horizon Diagnostics
How a research dashboard recorded each candidate’s first failing gate, then a frozen multi-horizon diagnostic confirmed no_trade before test access or model training.