Forecasting FinTech stock index under multiple market uncertainties
Researchers present the CPO-VMD-PConv-Informer model for forecasting the KBW Nasdaq Financial Technology Index (KFTX) amid economic and geopolitical risks. The model achieves high accuracy, with $R^2$ values of 0.9681 and 0.9757, and mean absolute percentage errors below 2 percent. SHAP analysis shows a shift in key predictors from term structure to Financial Stress Index after COVID-19. The study sets a new standard but notes the need for broader validation and discussion of limitations.
What it examines
This paper introduces a new CPO-VMD-PConv-Informer framework to forecast the KBW Nasdaq Financial Technology Index (KFTX) by considering eight key market uncertainty indicators, such as economic policy and geopolitical risks. The study aims to improve prediction accuracy during volatile and uncertain market conditions.
What it concludes
The proposed model outperforms traditional methods, maintaining high accuracy even during market shocks like COVID-19. Its robustness makes it valuable for investors, risk managers, and policymakers. Future research could expand to other indices or uncertainty factors, though limitations include reliance on selected indicators and data splits.
Evidence objects
A new forecasting model, CPO-VMD-PConv-Informer, predicts the KBW Nasdaq Financial Technology Index (KFTX) with exceptional accuracy, achieving $R^2$ values of 0.9681 and 0.9757, surpassing traditional machine learning methods.
key_findings bullet 1 · key_findings · validation V0
The models robustness stands out during extreme market events like COVID-19, thanks to its integration of Variational Mode Decomposition (VMD) and Crested Porcupine Optimizer (CPO), maintaining mean absolute percentage errors (MAPE) below 2%.
key_findings bullet 2 · key_findings · validation V0
SHAP analysis reveals a striking shift in predictive factors: before COVID-19, the term structure of the market (TM) dominated, but after the pandemic, the Financial Stress Index (FSI) became the main driver.
key_findings bullet 3 · key_findings · validation V0
This paper presents the CPO-VMD-PConv-Informer framework, uniquely integrating eight uncertainty indicators (including EPU, GPR) for KFTX forecasting. Combining VMD, CPO, and Informer, it outperforms traditional models, especially during extreme volatility (e.g., COVID-19). SHAP-based interpretability and robust empirical validation make it a compelling, innovative advancement in quantitative finance.
key_findings bullet 4 · key_findings · validation V0
Raw abstract and provenance
- … , Informer, to reduce computational complexity and improve forecast efficiency, particularly in the face of financial market shocks and macroeconomic uncertainties (Ren et …
Source row: 895 · abstract type: snippet