The Limits of Complexity: Why Feature Engineering Beats Deep Learning in Investor Flow Prediction

TL;DR

This study demonstrates that in financial prediction, simple feature engineering often outperforms complex deep learning models, especially in low signal-to-noise environments, highlighting the importance of domain knowledge.

Contribution

The paper provides empirical evidence that traditional feature engineering can surpass advanced deep learning techniques in investor flow prediction tasks.

Findings

Simple linear models with domain-specific features outperform complex models in predictive accuracy.

Deep learning models like LSTM with attention mechanisms underperform in noisy financial data.

Complex signal extraction methods do not significantly improve prediction over basic feature engineering.

Abstract

The application of machine learning to financial prediction has accelerated dramatically, yet the conditions under which complex models outperform simple alternatives remain poorly understood. This paper investigates whether advanced signal processing and deep learning techniques can extract predictive value from investor order flows beyond what simple feature engineering achieves. Using a comprehensive dataset of 2.79 million observations spanning 2,439 Korean equities from 2020--2024, we apply three methodologies: \textit{Independent Component Analysis} (ICA) to recover latent market drivers, \textit{Wavelet Coherence} analysis to characterize multi-scale correlation structure, and \textit{Long Short-Term Memory} (LSTM) networks with attention mechanisms for non-linear prediction. Our results reveal a striking finding: a parsimonious linear model using market capitalization-normalized flows (``Matched Filter'' preprocessing) achieves a Sharpe ratio of 1.30 and cumulative return of 272.6\%, while the full ICA-Wavelet-LSTM pipeline generates a Sharpe ratio of only 0.07 with a cumulative return of $-5.1\%$. The raw LSTM model collapsed to predicting the unconditional mean, achieving a hit rate of 47.5\% -- worse than random. We conclude that in low signal-to-noise financial environments, domain-specific feature engineering yields substantially higher marginal returns than algorithmic complexity. These findings establish important boundary conditions for the application of deep learning to financial prediction.