Paper: arXiv 2305.14382

Authors: Yuze Lu, Hailong Zhang, Qiwen Guo

Abstract

Applications of deep learning in financial market prediction has attracted huge attention from investors and researchers. In particular, intra-day prediction at the minute scale, the dramatically fluctuating volume and stock prices within short time periods have posed a great challenge for the convergence of networks result. Informer is a more novel network, improved on Transformer with smaller computational complexity, longer prediction length and global time stamp features. We have designed three experiments to compare Informer with the commonly used networks LSTM, Transformer and BERT on 1-minute and 5-minute frequencies for four different stocks/ market indices. The prediction results are measured by three evaluation criteria: MAE, RMSE and MAPE. Informer has obtained best performance among all the networks on every dataset. Network without the global time stamp mechanism has significantly lower prediction effect compared to the complete Informer; it is evident that this mechanism grants the time series to the characteristics and substantially improves the prediction accuracy of the networks. Finally, transfer learning capability experiment is conducted, Informer also achieves a good performance. Informer has good robustness and improved performance in market prediction, which can be exactly adapted to real trading.

Complexity vs Empirical Score

  • Math Complexity: 6.0/10
  • Empirical Rigor: 7.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper introduces advanced deep learning architectures (Informer, ProbSparse self-attention) with significant mathematical novelty, yet it also conducts a structured empirical comparison using standard financial metrics (MAE, RMSE, MAPE) on real intraday stock/index data, making it both theoretically dense and implementation-focused.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Intra-day Stock Prediction"] --> B["Data Preparation<br>1 & 5-min Frequencies"]
  B --> C{"Model Comparison"}
  C --> D["Informer<br>with Time Stamps"]
  C --> E["LSTM"]
  C --> F["Transformer"]
  C --> G["BERT"]
  
  D --> H["Performance Evaluation<br>MAE, RMSE, MAPE"]
  E --> H
  F --> H
  G --> H
  
  H --> I["Key Findings<br>Informer: Best Performance<br>Time Stamps: Critical<br>Transfer Learning: Effective"]

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 using the Informer network for minute‑level stock and market index prediction, highlighting its global time stamp mechanism.

Source sentence
Necessarily, in order to solve the above problems, in this paper we intro-duce Informer [38] into the intraday stock price prediction task.

Assumes 2

  • The closing price is used as the prediction target while other price and volume fields are treated as features. data
    Source sentence
    The closing price is defined as the predicted target, and the rest of data labels are training features.
  • 70% of the data is used for training and 10% for validation, implying a standard train‑validation split without considering transaction costs or market impact. execution
    Source sentence
    Among the whole data set, 70% of the data is set as training set to train the initial network; 10% of the data is set as validation…

Data

  • Hong Kong Hang Seng Index (HSI) market-prices unknown
    1‑minute: 03 January 2022 to 12 December
    Source sentence
    (IXIC), Tencent Holdings Ltd and Apple Inc. (AAPL). The data involved in the training includes opening, closing, high and low prices, volume and turnover ratio at 1‑minute and 5‑minute frequencies.
  • NASDAQ Index (IXIC) market-prices unknown
    1‑minute: 03 January 2022 to 12 December
    Source sentence
    In this work, two representative stock price indices and two typical company stocks are selected: Hong Kong Hang Seng Index (HSI), NASDAQ Index
  • Tencent Holdings Ltd market-prices unknown
    1‑minute: 03 January 2022 to 12 December
    Source sentence
    In this work, two representative stock price indices and two typical company stocks are selected:
  • Apple Inc. (AAPL) market-prices unknown
    1‑minute: 03 January 2022 to 12 December
    Source sentence
    In this work, two representative stock price indices and two typical company stocks are selected:

Costs

Not applicable The paper reports no trading or execution result.

Finds 4

  • Informer achieves the best performance across all datasets and evaluation metrics (MAE, RMSE, MAPE) compared with LSTM, Transformer and BERT. performance
    Source sentence
    Informer has obtained best performance among all the networks on every dataset.
  • The global time stamp mechanism significantly improves prediction accuracy; removing it leads to notably lower performance. predictive-accuracy
    Source sentence
    Network without the global time stamp mechanism has significantly lower prediction effect compared to the complete Informer.
  • Informer maintains superior performance even on highly volatile stock price periods. performance
    Source sentence
    Informer has the best performance not only when the stock price is stable but when it is highly volatile.
  • Transfer learning with a pre‑trained Informer (InformerT) still yields good performance, though slightly worse than the full Informer. performance
    Source sentence
    Table 5 Evaluation criteria for Informer and pre-trained InformerT

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 39 related papers). Coverage is partial, so this is neither replication nor refutation.

Cited by 1 paper in the archive (1 by independent authors): Informer in Algorithmic Investment Strategies on High … (used as a baseline)

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.