Backtesting, done the way it has to be done
Written by the Capstone Metals research desk. Reviewed by Daniel Kenney, President. Last updated February 2026.
The rules our harness enforces
- The harness owns the clock. A strategy is handed history up to a date and never sees past it.
- Costs are mandatory. Every change in position is charged 8 basis points of fees and 12 basis points of assumed half-spread, and a stress run multiplies slippage by 3.
- Exposure is capped. Position and gross-exposure limits are applied by the harness, not left to the strategy's good manners.
- The holdout is cut first. A slice of history is reserved and locked before any result is produced, and tested once.
- Nothing is judged by a model. Every return, ratio, drawdown and cost figure is computed by ordinary deterministic code. No language model calculates a number on this site.
Six ways a backtest lies
Look-ahead
Using information that was not available when the decision was supposedly made. Our harness hands a strategy only the bars dated on or before the decision date, and applies the result to the next period — the strategy is never given the option of cheating.
Survivorship
Testing only the instruments that still exist today. A universe built from today's winners will flatter any rule applied to it.
Costless trading
A rule that trades often can look excellent until fees and spread are charged. Our harness cannot express a zero-cost run; costs are part of the interface, not an option.
Revision blindness
Backtesting against numbers that were later corrected. Our price history is stored with publication vintages, so a later revision is invisible to an earlier test.
Quiet re-testing
Trying variants until something passes. The holdout window is reserved and locked before the first backtest, trials are counted, and a re-tuned idea becomes a new version rather than a better score for the old one.
Ignoring the alternative
Reporting a return without asking what simply holding the same assets would have done. Every run here carries its buy-and-hold baseline.
The history we test against
The classified history is still being assembled. We would rather show nothing than describe a record we do not hold.
What a passed backtest earns
A simulated order in paper trading, after a named human approves it. Not real money. Live execution is not built into this system, and turning it on would take a deliberate build plus independent risk approval — not a setting.
Evidence and sources
- 1. Bailey & López de Prado, The Deflated Sharpe Ratio (2014) — Why a Sharpe ratio must be discounted by the number of trials that produced it.
- 2. Harvey, Liu & Zhu, …and the Cross-Section of Expected Returns (2016) — The multiple-testing problem behind most published factors.
- 3. Moskowitz, Ooi & Pedersen, Time Series Momentum (2012) — A widely replicated result, and the kind of evidence we treat as a hypothesis rather than a conclusion.
Where this fits in the wider picture
Market Intelligence is the evidence layer. The wealth-protection side of Capstone is where those findings meet an actual plan — metals, retirement accounts and stewardship of what you already hold.
- The education library — money, debt and purchasing power, from first principles.
- Gold's place in the world economy — what the historical record does and does not show.
- Wealth protection in one place — how metals, advice and entities fit together.
Research and education only. Nothing on this page is investment advice, a recommendation to buy or sell any security or metal, or a forecast. No outcome is promised or implied. Simulated and historical results do not indicate future results, and any strategy discussed here may lose money. Speak with us about your own circumstances before acting: (800) 200-9553.