Paper: arXiv 2304.09840

Authors: Adamantios Ntakaris, Moncef Gabbouj, Juho Kanniainen

Abstract

High-frequency trading requires fast data processing without information lags for precise stock price forecasting. This high-paced stock price forecasting is usually based on vectors that need to be treated as sequential and time-independent signals due to the time irregularities that are inherent in high-frequency trading. A well-documented and tested method that considers these time-irregularities is a type of recurrent neural network, named long short-term memory neural network. This type of neural network is formed based on cells that perform sequential and stale calculations via gates and states without knowing whether their order, within the cell, is optimal. In this paper, we propose a revised and real-time adjusted long short-term memory cell that selects the best gate or state as its final output. Our cell is running under a shallow topology, has a minimal look-back period, and is trained online. This revised cell achieves lower forecasting error compared to other recurrent neural networks for online high-frequency trading forecasting tasks such as the limit order book mid-price prediction as it has been tested on two high-liquid US and two less-liquid Nordic stocks.

Complexity vs Empirical Score

  • Math Complexity: 5.5/10
  • Empirical Rigor: 6.5/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper presents a novel structural modification to standard LSTM cells for online high-frequency trading tasks, specifically targeting irregular time series in limit order books. The empirical evaluation spans multiple liquidity regimes (US and Nordic equities) and demonstrates improved forecasting accuracy, though the detailed mathematical derivations and reproducible codebase details are limited in the summary.

Research Flowchart

  flowchart TD
    A[Goal: Optimize LSTM Cell for Real-Time HFT Price Forecasting] --> B[Data Input: LOB Mid-Prices from US & Nordic Equities]
    B --> C[Method: Shallow Topology Online Training with Minimal Look-Back]
    C --> D[Model Process: Real-Time Selection of Optimal Gate/State Output]
    D --> E[Outcome: Lower Forecasting Error vs Standard RNNs in HFT]