Walk-forward evaluation: for each quarter t, the model uses only data up to t-1 (no lookahead) to predict t. Residuals and error metrics measure how well the deterministic engine would have anticipated the actual reading.
| Index | N | MAE | RMSE | MAPE | Bias | Hits ±5 |
|---|---|---|---|---|---|---|
| CMHI · Market Health | 22 | 1.93 | 2.83 | 3.5% | +0.07 | 86% |
| CVI · Vulnerability | 22 | 2.27 | 3.39 | 3.3% | -0.13 | 82% |
| CBTI · Trust | 22 | 0.95 | 1.53 | 2.0% | +0.05 | 100% |
| MRI · Market Risk | 22 | 2.81 | 4.31 | 4.2% | -0.17 | 77% |
| CERI · Resilience | 22 | 1.07 | 1.64 | 3.1% | +0.09 | 100% |
| NCES · Composite | 22 | 2.00 | 3.08 | 4.3% | +0.11 | 86% |
Solid line = actual reading; dashed green = walk-forward prediction; orange area = residual (actual − predicted). Lower MAE/RMSE and a bias near zero indicate an unbiased forecast.
| Quarter | Actual | Predicted | Residual |
|---|---|---|---|
| 2021 Q1 | 51.50 | 47.10 | +4.40 |
| 2021 Q2 | 52.60 | 53.15 | -0.55 |
| 2021 Q3 | 53.50 | 54.05 | -0.55 |
| 2021 Q4 | 54.10 | 54.75 | -0.65 |
| 2022 Q1 | 52.10 | 55.10 | -3.00 |
| 2022 Q2 | 46.90 | 52.85 | -5.95 |
| 2022 Q3 | 41.40 | 46.20 | -4.80 |
| 2022 Q4 | 39.20 | 37.80 | +1.40 |
| 2023 Q1 | 41.80 | 33.85 | +7.95 |
| 2023 Q2 | 44.70 | 37.95 | +6.75 |
| 2023 Q3 | 47.20 | 44.90 | +2.30 |
| 2023 Q4 | 49.00 | 49.95 | -0.95 |
| 2024 Q1 | 50.50 | 51.70 | -1.20 |
| 2024 Q2 | 51.70 | 52.65 | -0.95 |
| 2024 Q3 | 52.70 | 53.35 | -0.65 |
| 2024 Q4 | 53.60 | 54.05 | -0.45 |
| 2025 Q1 | 54.80 | 54.70 | +0.10 |
| 2025 Q2 | 55.60 | 55.75 | -0.15 |
| 2025 Q3 | 56.20 | 56.65 | -0.45 |
| 2025 Q4 | 56.70 | 57.20 | -0.50 |
| 2026 Q1 | 57.50 | 57.40 | +0.10 |
| 2026 Q2 | 58.20 | 58.05 | +0.15 |