Table 3.
In-sample model fit comparison criteria
| Criteria | MAE × 102 | |||||
|---|---|---|---|---|---|---|
| Period | Pre-COVID-19 | COVID-19 | ||||
| Model | LMM | GAM | Tv-LMM | LMM | GAM | Tv-LMM |
| BPI | 1.026 | 0.995 | 0.693 | 0.977 | 0.947 | 0.498 |
| ETH | 1.056 | 1.047 | 0.710 | 1.061 | 1.017 | 0.594 |
| XRP | 1.231 | 1.231 | 0.765 | 1.003 | 1.004 | 0.776 |
| BCH | 1.631 | 1.560 | 1.162 | 1.267 | 1.228 | 1.008 |
| BNB | 2.171 | 2.162 | 1.996 | 1.623 | 1.606 | 1.599 |
| LINK | 3.843 | 3.818 | 3.694 | 3.006 | 2.997 | 2.387 |
| CRO | 3.602 | 3.602 | 3.600 | 1.777 | 1.772 | 1.654 |
| LTC | 1.580 | 1.573 | 1.321 | 1.141 | 1.101 | 0.910 |
| BSV | 3.073 | 2.844 | 2.145 | 1.614 | 1.521 | 0.783 |
| ADA | 1.651 | 1.640 | 1.517 | 1.920 | 1.911 | 1.810 |
| Average | 2.086 | 2.047 | 1.760 | 1.539 | 1.510 | 1.202 |
| Criteria | MSE × 104 | |||||
|---|---|---|---|---|---|---|
| Period | Pre-COVID-19 | COVID-19 | ||||
| Model | LMM | GAM | Tv-LMM | LMM | GAM | Tv-LMM |
| BPI | 2.373 | 2.245 | 1.058 | 1.748 | 1.692 | 0.476 |
| ETH | 2.532 | 2.485 | 1.134 | 2.163 | 2.003 | 0.694 |
| XRP | 3.793 | 3.689 | 1.330 | 2.291 | 2.291 | 1.414 |
| BCH | 6.765 | 6.140 | 3.410 | 3.135 | 2.944 | 1.979 |
| BNB | 8.923 | 8.756 | 7.625 | 6.151 | 6.071 | 6.030 |
| LINK | 31.317 | 30.971 | 29.183 | 18.686 | 18.521 | 11.819 |
| CRO | 51.187 | 51.187 | 51.180 | 6.543 | 6.511 | 5.641 |
| LTC | 5.296 | 5.260 | 3.445 | 2.618 | 2.433 | 1.597 |
| BSV | 38.860 | 30.440 | 18.874 | 5.975 | 5.375 | 1.322 |
| ADA | 5.181 | 5.001 | 4.300 | 8.538 | 8.468 | 7.484 |
| Average | 15.623 | 14.617 | 12.154 | 5.785 | 5.631 | 3.846 |
Bold values indicate the best modeling performance for cryptocurrency price in terms of the lowest MAE and MSE, respectively