This study presents a novel hybrid decision-support framework that integrates financial signal processing (FSP) techniques with a multi-model machine learning (ML) validation layer to minimize the risks posed by high volatility and false signals in crypto-asset markets. Unlike traditional forecasting models, the proposed system does not employ the predictive component as a standalone signal generator; instead, it is positioned as a Binary Confirmation Mechanism (BCM) that validates technical data. The research methodology consists of four fundamental stages: First, the parameters of the Super Trend (ST), Williams %R (WR), and Williams Fractal (WF) indicators are optimized using 5-fold Time-Series Cross-Validation on BTC-USD data spanning from 2014 to 2021. Second, nine different hybrid signal scenarios including single, dual and full combinations are derived from these optimized indicators. Third, the generated signals are filtered through eight distinct ML algorithms, including Random Forest (RF), Support Vector Machines (SVM), and Multi-Layer Perceptron (MLP). In this stage, a buy-side execution requires a match between the technical signal and ML approval, whereas an asymmetric priority is given to technical signals for sell-side operations to ensure rapid exit and capital preservation. Finally, the system's performance is evaluated through portfolio simulations on out-of-sample test data from 2021 to 2026. The findings demonstrate that the developed hybrid approach successfully mitigates false signals, elevates the cumulative equity curve above the Buy & Hold benchmark, and provides more stable risk management by reducing maximum drawdowns. This longitudinal analysis covering a 12-year period reveals that validating optimized financial signals with artificial intelligence significantly enhances execution consistency in algorithmic trading strategies.
Seçkin Karasu (Tue,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: