Why normalize?

Valuation distance, volatility, momentum and drawdown all use different units. Normalization maps components to a common range so they can be combined without one dominating merely because of its numerical scale.

Weights

Weights express model design choices. obsila Market Risk v1.1 assigns the greatest weight to valuation position, then combines trend extension, volatility, drawdown recovery and momentum. The exact weights are documented in Methodology.

Historical calibration

Rather than assigning labels solely from arbitrary fixed thresholds, obsila compares the score with its own historical distribution. Percentile-based bands help answer whether today's score is low or high relative to the model's past.

Backtesting traps

Overfitting, look-ahead bias, survivorship bias and repeated parameter tuning can make historical results appear more useful than they really are. A credible model therefore needs versioning and documented assumptions.

A transparent quantitative diagnostic can improve consistency. It cannot remove uncertainty.