The gold standard for testing financial models is fundamentally broken and mathematically impossible for a new class of physics-based trading systems.
April 26, 2026
Original Paper
Ontological Limits of Historical Validation for Wavefunction-Calibrated Financial Systems
SSRN · 6614899
The Takeaway
Classical backtesting is structurally incapable of validating systems that use wavefunction-based models due to a phenomenon called dimensional collapse. These advanced AI systems operate in a way that makes historical data a poor predictor of future outcomes. Most banks and hedge funds still rely on backtesting to prove their models are safe and effective. This discovery proves that for the most sophisticated systems, that proof is a logical impossibility. This realization will force the entire financial industry to find new ways to verify the safety of automated trading.
From the abstract
We establish that classical backtesting is structurally incompatible with wavefunction-calibrated financial systems through three independent arguments. The dimensional collapse problem shows that the options-implied wavefunction ψ(S,t)-a field over the entire price domain-cannot be reconstructed from the realized price trajectory or from aggregate proxies such as VIX. Ergodicity breaking shows that time-averaged evaluation is circular for a system designed for non-ergodic multiplicative dynamic