Paper: arXiv 2310.00747

Authors: Hsiang-Hui Liu, Han-Jay Shu, Wei-Ning Chiu

Abstract

We introduce NoxTrader, a sophisticated system designed for portfolio construction and trading execution with the primary objective of achieving profitable outcomes in the stock market, specifically aiming to generate moderate to long-term profits. The underlying learning process of NoxTrader is rooted in the assimilation of valuable insights derived from historical trading data, particularly focusing on time-series analysis due to the nature of the dataset employed. In our approach, we utilize price and volume data of US stock market for feature engineering to generate effective features, including Return Momentum, Week Price Momentum, and Month Price Momentum. We choose the Long Short-Term Memory (LSTM)model to capture continuous price trends and implement dynamic model updates during the trading execution process, enabling the model to continuously adapt to the current market trends. Notably, we have developed a comprehensive trading backtesting system - NoxTrader, which allows us to manage portfolios based on predictive scores and utilize custom evaluation metrics to conduct a thorough assessment of our trading performance. Our rigorous feature engineering and careful selection of prediction targets enable us to generate prediction data with an impressive correlation range between 0.65 and 0.75. Finally, we monitor the dispersion of our prediction data and perform a comparative analysis against actual market data. Through the use of filtering techniques, we improved the initial -60% investment return to 325%.

Complexity vs Empirical Score

  • Math Complexity: 3.0/10
  • Empirical Rigor: 6.0/10
  • Quadrant: Street Traders — empirical and applied, lower math complexity

Why this score: The paper uses standard LSTM and basic statistical metrics without complex derivations, but presents a full backtesting system with concrete performance results (325% return) and detailed feature engineering from real US stock data.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Predict Stock Return Momentum<br>for Profitable Quantitative Trading"] --> B["Data & Feature Engineering<br>US Stock Price & Volume Data<br>Features: Return Momentum,<br>Week/Month Price Momentum"]
  B --> C["LSTM Model Training<br>Captures Continuous Price Trends<br>Dynamic Updates for Adaptation"]
  C --> D["Trading Backtesting System<br>NoxTrader: Portfolio Construction<br>& Execution based on Predictions"]
  D --> E["Key Findings & Outcomes<br>Prediction Correlation: 0.65-0.75<br>Filtered Returns: -60% to 325%"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes NoxTrader, an LSTM‑based system for stock return momentum prediction, portfolio construction and trading execution.

Source sentence
We introduce NoxTrader, a sophisticated system designed for portfolio construction and trading execution with the primary objective of achieving profitable outcomes in the stock market, specifically aiming to generate moderate to long‑term profits.

Assumes 4

  • Features are generated from US companies with market capitalization greater than 200 billion dollars. data
    Source sentence
    We collect market data for US companies with a market capitalization of more than 200 billion dollars and draw upon a rich spectrum of insights that collectively enhance the model's ability.
  • The model is retrained every 10 days to remain adaptive to evolving market patterns. model
    Source sentence
    Importantly, considering the temporal nature of stock data, retraining the model is necessary every 10 days to ensure its adaptability to evolving market patterns.
  • Prediction horizon is limited to 10‑day windows because correlation declines for longer horizons. model
    Source sentence
    While a predictive horizon beyond 10 days might be appealing, we observed a decline in correlation beyond this point.
  • Labels are defined as return momentum (difference between returns of two consecutive days). data
    Source sentence
    The labels represent the difference between return of two consecutive days, namely return momentum.

Data

Backtest sample: US stock market · equities

  • yfinance historical price data · yfinance market-prices public
    US stocks (market cap >200 B)
    Source sentence
    The historical data supplied by the yfinance platform embodies a set of fundamental raw metrics, encompassing the opening, closing, highest, and lowest values of each individual stocks.

Recognised sources: Yahoo Finance

Costs

Partial Costs appear only as a robustness check, a flat fee without spread or impact, or outside the headline numbers.

Finds 4

Reported performance (NoxTrader): total return 325.0%

  • Prediction data achieve a correlation range between 0.65 and 0.75. predictive-accuracy
    Source sentence
    Our rigorous feature engineering and careful selection of prediction targets enable us to generate prediction data with an impressive correlation range between 0.65 and 0.75.
  • Filtering techniques improved the initial –60% investment return to a 325% return. performance
    Source sentence
    Through the use of filtering techniques, we improved the initial -60% investment return to 325%.
  • Switching label representation to return momentum increased correlation to 0.6. predictive-accuracy
    Source sentence
    This alteration yielded remarkable results as the correlation coefficient surged to an impressive 0.6…
  • Earlier correlation between predicted and true return differentials was modest at 0.2. predictive-accuracy
    Source sentence
    yielding an improved but modest correlation coefficient of 0.2.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 41 related papers). Coverage is partial, so this is neither replication nor refutation.

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.