ADAM INFERENCE ENGINE | ON-DEVICE ML: ACTIVE
LATENCY: <1ms

Deterministic Quant Module

Open-Weight Structural Financial Models

Entity Features (X)

Leverage (Debt/EBITDA) 3.5x

Weight: +0.85 (Increases PD)

Interest Coverage Ratio 4.0x

Weight: -0.45 (Decreases PD)

Operating Margin (%) 15.0%

Weight: -0.05 (Decreases PD)

Asset Tangibility (%) 40%

Drives Recovery Rate (Inverse of LGD)

Deterministic ML: Logistic Regression

The Probability of Default (PD) is calculated via a sigmoid activation applied to the linear combination of features and open weights:

z = β₀ + (β₁·Lev) + (β₂·ICR) + (β₃·Marg)
PD = 1 / (1 + e⁻ᶻ)
Open Weights Matrix (W)
Bias (β₀): -2.500
Leverage (β₁): +0.850
Coverage (β₂): -0.450
Margin (β₃): -0.050

Probability of Default (1Y)

4.52%
IMPLIED RATING: BB-

Loss Given Default (LGD)

60.0%
Recovery Rate: 40.0%

LGD = 1 - (Base Recovery + Tangibility Factor)

Feature Importance (Log-Odds Contribution)