Paper: arXiv 2407.09831

Authors: Han Gui

Abstract

To predict the future movements of stock markets, numerous studies concentrate on daily data and employ various machine learning (ML) models as benchmarks that often vary and lack standardization across different research works. This paper tries to solve the problem from a fresh standpoint by aiming to predict the weekly movements, and introducing a novel benchmark of random traders. This benchmark is independent of any ML model, thus making it more objective and potentially serving as a commonly recognized standard. During training process, apart from the basic features such as technical indicators, scaling laws and directional changes are introduced as additional features, furthermore, the training datasets are also adjusted by assigning varying weights to different samples, the weighting approach allows the models to emphasize specific samples. On back-testing, several trained models show good performance, with the multi-layer perception (MLP) demonstrating stability and robustness across extensive and comprehensive data that include upward, downward and cyclic trends. The unique perspective of this work that focuses on weekly movements, incorporates new features and creates an objective benchmark, contributes to the existing literature on stock market prediction.

Complexity vs Empirical Score

  • Math Complexity: 2.0/10
  • Empirical Rigor: 4.0/10
  • Quadrant: Philosophers — conceptual discussion, lower math and empirical scores

Why this score: The paper employs standard ML techniques (e.g., MLP) and basic financial metrics without advanced mathematical derivations, but its empirical focus on weekly predictions, specific features (scaling laws, directional changes), and a proposed ‘random trader’ benchmark indicates a practical, implementation-heavy approach, albeit likely limited by the absence of detailed code or backtest metrics in the excerpt.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Predict Weekly Stock Movements"] --> B["Novel Benchmark<br>Random Trader"]
  A --> C["Methodology<br>Scaling Laws & Directional Changes"]
  B --> D["Data & Models<br>Weighted Datasets + MLP/LSTM/GBM"]
  C --> D
  D --> E["Back-Testing<br>Rolling Window Validation"]
  E --> F{"Outcomes"}
  F --> G["MLP: Robust & Stable<br>Across Up/Down/Cyclic Trends"]
  F --> H["New Objective Standard<br>Weekly Prediction Focus"]

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 predicting weekly stock market movements and introduces a novel benchmark of random traders to objectively evaluate model performance.

Source sentence
This paper tries to solve the problem from a fresh standpoint by aiming to predict the weekly movements, and introducing a novel benchmark of random traders.

Assumes 4

  • Trading results are evaluated without accounting for market frictions or impacts. execution
    Source sentence
    it does not account for market frictions and market impacts that can influence trading outcomes.
  • Short‑selling is not considered in the trading strategy. execution
    Source sentence
    the work does not consider the short‑selling mechanism for trading strategies
  • Training samples are weighted according to their return magnitude, implying that larger moves are more important for profitability. model
    Source sentence
    training dataset is augmented by giving more duplicates to samples with bigger changes, here the duplicates do not imply over‑fitting, they give that sample more weights in the training
  • Weekly predictions assume that using current week information can forecast the next week’s direction. data
    Source sentence
    use current week information to predict next week movement

Data

Recognised sources: Yahoo Finance

Costs

Not modeled A trading or execution result is reported without costs.

Source sentence
it does not account for market frictions and market impacts that can influence trading outcomes.

Finds 3

  • On back‑testing, several trained models show good performance, with the multi‑layer perceptron (MLP) demonstrating stability and robustness across extensive and comprehensive data. performance
    Source sentence
    On back‑testing, several trained models show good performance, with the multi‑layer perception (MLP) demonstrating stability and robustness across extensive and comprehensive data
  • Higher accuracy of random traders generally leads to higher expected gross returns, establishing the benchmark as an objective, model‑independent measure. performance
    Source sentence
    Higher accuracy leads to higher profitability when calculating the expected return of cases with the same accuracy such as the cases with 75% accuracy generally have higher returns than that with 50% accuracy.
  • Random traders with 50% accuracy produce an average return close to market performance, while accuracies of 40% can achieve positive profit for all indexes and most stocks. performance
    Source sentence
    It shows that accuracy=40% can achieve positive profit for all indexes and most of stocks, while accuracy=60% can guarantee a profit for all indexes and stocks.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 11 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.