Illustrative mathematical surface · Research note
Why a compelling equity curve is only the beginning. A practical approach to testing whether a signal survives. This illustrative research note sets out an approach, its assumptions, and the questions left open.
Research
repositories
Get the notebooks, modeling code, and supporting datasets behind the videos as they become available.
- Full research archive
- New published projects
- Reproducible experiments
Membership opens when the first repository files are ready. No payments are being collected.
Recurring monthly subscription · Cancel anytimeA good-looking line is a hypothesis
A backtest summarizes decisions made under a set of assumptions. It cannot tell us whether those assumptions would have held in a real trading environment. This note outlines checks that should come before confidence.
Keep time intact
Split observations chronologically. Fit transformations on training data only, and carry those fitted transformations into validation. If labels overlap in time, leave an appropriate gap between splits.
split = int(len(observations) * 0.7)
train = observations[:split]
test = observations[split:]
The snippet is pseudocode for the split, not a complete trading implementation.
Methodology
Start with a simple baseline. Record every strategy variant considered. Add transaction costs, turnover, and capacity assumptions explicitly. Stress test the result under different market periods, and compare it with a suitable benchmark.
Interpreting results
The strongest question is whether a proposed mechanism survives reasonable changes to the assumptions. A result that disappears under small cost changes deserves a different level of confidence than one with a large margin.
Limitations
This is an illustrative testing protocol. It provides no equity returns, performance claims, or investment recommendations. Survivorship bias and historical data revisions require dedicated treatment.
