A Controlled Comparison of Deep Learning Architectures for Multi-Horizon Financial Forecasting: Evidence from 918 Experiments

By Nabeel Ahmad Saidd

Published 2026-02-27

Everscope rating

1762.9

Relevance to quantitative trading

8 / 10

Implementation complexity

6 / 10

Reproducibility

5 / 5

About this paper

Methodology: Five-Stage Controlled Benchmarking Protocol. Problem types: Time Series Forecasting, Regression, Ranking, Risk Management, Portfolio Optimization.

arXiv:2603.16886 · Code · Paper rankings

Open the interactive Everscope explorer for full analysis, charts, and paper battles.