Case studies

Framework validation against real crises

Every score, every PBO, every forward result is published.

Latest11 min read

The Pandemic Stress Test: 25 Strategies Through COVID-19

0.706 Spearman rho: strongest forward correlation across all five events

The COVID crash was faster and more violent than 2008, but the recovery was also faster. We ran 25 strategies through validation and achieved a 0.706 Spearman rank correlation between composite scores and forward performance.

framework validationCOVID-19forward validationcrisis testing
All case studies
01

The Pandemic Stress Test: 25 Strategies Through COVID-19

0.706 Spearman rho: strongest forward correlation across all five events
11 min read
02

24 Strategies Walk Into a Financial Crisis

45.7-point separation between good and bad strategy composites
12 min read
03

The Day Volatility Exploded: 12 Strategies Through Volmageddon

Framework discrimination held (27.1-point gap) despite the toughest event in the suite
10 min read

Methodology

How we test the framework

01

Train on pre-crisis data

Strategies are trained and optimized on data that ends before the crisis begins. For the GFC study, training uses 2004 to 2007 data.

02

Validate against the crisis

Run the full seven-layer pipeline against the crisis period. Measure composite scores, PBO, and crisis Sharpe ratios.

03

Measure forward performance

Compare validation scores to actual post-crisis returns. Spearman rank correlation tells us if rankings predict real outcomes.

Results

Key findings across all studies

45.7
Max discrimination gap (good vs bad)
0.706
Max forward correlation (Spearman rho)
61
Total strategies tested
30+
Years of crisis data covered

The forward correlation means that for roughly three-quarters of strategy pairs, the validation ranking correctly predicted which strategy would outperform in the years following each crisis.