Foundation group
Research Design
The controls that separate an interesting backtest from credible evidence.
Guide map
Read only as far as the decision requires.
Each guide stands on its own while linking forward to the strategy or Results question it supports.
Research Design guides
What a backtest can and cannot show
A backtest is a historical model of declared rules and assumptions—not a forecast or a guarantee.
Data integrity: survivorship, point-in-time inputs, and missing data
A clean-looking dataset can still contain future knowledge, missing failures, or silent substitutions.
Baselines, costs, and fair comparisons
A strategy earns attention only after it is compared on the same universe, timing, risk, and cost basis as a credible alternative.
Holdout and walk-forward testing
Evaluate a fixed research decision on later data while preserving the order in which information became available.
Overfitting, multiple testing, and parameter search
The more ideas and settings you try, the easier it becomes to discover a persuasive result that is only noise.