Placebo Arm
Methodology· DE: Placebo-Arm
A comparison built to have no edge by construction (random entry days, shuffled signals) and run through the same pipeline as the claim. If the claim does not beat the spread of its placebo, the claim falls — whatever its raw number. A placebo counts only if it was built before the analysis.
Calculation
placebo = { built_before_analysis, method, p_low, p_high } — the claim's value is read against the placebo distribution's quantiles.
Unit & source
mixed · tradingstrategies.work falsification register (spec-falsifikations-register-v1)
Related terms
Definitions are for research and education. Metrics describe market conditions — not financial advice or a buy/sell signal.