EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Dynamic Stability and Neural Residual Forecasting of an Inverse Lomax Autoregressive Model with Application to NSPO Stock Prices

Aida A. Mohammed

Modern Journal of Statistics · 15.11.2026 · 2 (2) · 237-257

DOI: 10.64389/mjs.2026.022133 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Modern Journal of Statistics, online erschienen am 15.11.2026. Verfasst von Aida A. Mohammed. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This study investigates an inverse-Lomax smooth-transition autoregressive (ILAR) model integrated with neural residual forecasting, using 319 weekly NSPO stock-price observations. The ILAR(1) map has a positive deterministic fixed point of 2.3507 and local spectral radius of 0.9345, but its derivative is 1.0193. The fitted deterministic map does not satisfy the sufficient global-contraction condition over the examined state interval. The estimated transition-shape parameter reaches its imposed upper bound ( ) and remains boundary-constrained when the bound is extended to 8. ADF and Phillips–Perron tests do not reject the unit-root null for price levels at 5%. The initial single-run ILAR-TCNA evaluation yielded an RMSE of 0.1427, compared with 0.1474 for the naive benchmark. A separate 20-seed experiment with trainable four-head attention produced mean test RMSE 0.1513 (SD 0.0063) and mean MAE 0.1215 (SD 0.0050). The validation procedure selected zero residual weight in 16 runs; the seed-4 Diebold–Mariano tests did not reject equal predictive accuracy against the naive and AR(1) forecasts. Penalized transition-parameter profiles also support weak identification. These findings indicate local deterministic stability but neither stochastic stationarity nor a reproducible neural forecasting advantage.

Zitieren

APA

Mohammed, A. A. (2026). Dynamic Stability and Neural Residual Forecasting of an Inverse Lomax Autoregressive Model with Application to NSPO Stock Prices. Modern Journal of Statistics, 2(2), 237-257. https://doi.org/10.64389/mjs.2026.022133

Vancouver

Mohammed AA. Dynamic Stability and Neural Residual Forecasting of an Inverse Lomax Autoregressive Model with Application to NSPO Stock Prices. Modern Journal of Statistics. 2026;2(2):237-257. doi:10.64389/mjs.2026.022133

BibTeX
@article{Mohammed2026,
  author = {Mohammed, Aida A.},
  title = {Dynamic Stability and Neural Residual Forecasting of an Inverse Lomax Autoregressive Model with Application to NSPO Stock Prices},
  journal = {Modern Journal of Statistics},
  year = {2026},
  volume = {2},
  number = {2},
  pages = {237--257},
  doi = {10.64389/mjs.2026.022133}
}
Kostenlos starten →