Latest / Weekly Working Paper Digest - Market Micro Structure/Retail Trading
Deep Dive: Explainable Deep Learning for Price-Trade Dynamics: From Black-Box Forecasts to Effective Parametric Models
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!
This paper demonstrates that deep neural networks capture genuine nonlinear mechanisms in high-frequency financial data, specifically saturating order-flow effects and state-dependent price reversals, which linear models miss. By using Shapley-based explainability to extract these functional forms, the authors construct a parsimonious parametric model that matches or exceeds the neural network's predictive performance while offering greater stability and interpretability. This bridges the…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.