All reviews of published articles are made public. This includes manuscript files, peer review comments, author rebuttals and revised materials. Note: This was optional for articles submitted before 13 February 2023.
Peer reviewers are encouraged (but not required) to provide their names to the authors when submitting their peer review. If they agree to provide their name, then their personal profile page will reflect a public acknowledgment that they performed a review (even if the article is rejected). If the article is accepted, then reviewers who provided their name will be associated with the article itself.
Good job, congratulations on the acceptance of your manuscript.
[# PeerJ Staff Note - this decision was reviewed and approved by Xiangjie Kong, a PeerJ Section Editor covering this Section #]
'No Comment'
'No Comment'
'No Comment'
Dear Authors,
Thank you for submitting the revised version of your manuscript. I have thoroughly reviewed the updates and greatly appreciate the effort you have invested in addressing all the feedback. The revisions have significantly enhanced the clarity and overall quality of the paper. Below, I provide my overall comments on the manuscript.
Overall Comments:
The manuscript now effectively presents the methodology, discussion, and results, with a clear explanation of the hybrid deep learning model (LSTM+RL). The addition of performance metrics and comparisons with traditional models has further strengthened the validity of your findings and the hybrid model (LSTM+RL).
The discussion section has been improved by incorporating insights into the practical applications of the model, particularly in real-world asset management scenarios. This anchors the technical work within its practical context.
Please revise the manuscript according to the review
Please, remember that it is PeerJ policy that additional references suggested during the peer-review process should only be included if the authors are in agreement that they are relevant and useful.
I greatly value the authors' contribution to the application of deep learning in financial management, a highly pertinent and expanding area of research. This study addresses a modern and relevant topic.
The manuscript is well-organized, featuring a clear introduction, a comprehensive literature review, and a detailed methodology, making it easy to follow the authors' argument. and effectively positions the research within the existing body of knowledge.
I appreciate the authors for supplying the raw data, spanning from July 1, 2010, to October 24, 2023, collected over a thirteen-year period for the selected top ten companies.
While the results section is well-structured with clear subsections, a more in-depth explanation of the proposed mechanism is needed. Referencing relevant research work that discusses traditional comparative analysis and its outcomes could strengthen more this section.
The results show a notable enhancement in predictive accuracy, supported by detailed tables and graphs that depict the performance metrics.
The study employs a hybrid deep learning model combining LSTM and RL techniques. However, the model's training process well defined with the detailed explanation of the parameter tuning, which is crucial for reproducibility.
However, I recommend that the authors explore the robustness of the proposed mechanism (LSTM and RL) by determining forecast accuracy measures such as MAE, MAPE, and RMSE for the training and validation sets of the top ten stock companies with the lowest error rates.
Additionally, I suggest evaluating the trading performance through metrics such as the Sharpe ratio, maximum drawdown, and information ratio for the given stocks.
The discussion emphasizes the practical applications of the hybrid model in real-world stock asset management but lacks a comprehensive comparison with the traditional ARIMA model and the non-linear MLP model.
Comparing the hybrid model's performance with these established benchmarks and addressing the interpretability of its predictions for financial decision-making would further strengthen the study.
Consider incorporating other technical indicators such as Williams R%, stochastic K% & D% to evaluate and compare the results of the model.
In this paper, LSTM and RL were combined for trading and prediction as a hybrid model by comprehending stock trends. The comprehension through LSTM assists in assets management with minimum risk.
Literature review is well written. However, the following reference can be added. Write-up is clear and professional throughout the paper.
QM Ilyas, K Iqbal, S Ijaz, A Mehmood, S BhatiaA hybrid model to predict stock closing price using novel features and a fully modified hodrick–Prescott filter,Electronics 11 (21), 3588
**PeerJ Staff Note:** It is PeerJ policy that additional references suggested during the peer-review process should only be included if the authors are in agreement that they are relevant and useful.
Figure 1 quality needs improvement.
Experiments were performed with Table 1 data having 10-well known American companies. However, threshold values are too general as Thetac1 and Thetac2 in equation 7 and equation 8 that needs further explanation to present it in a concise manner in a table for quick view for readers rather than on Page 12. Particularly, the way of setting threshold through experimentation should be clearly mentioned in the paper while taking difference between time steps t and t-1.
Figures and Tables are ok Except Figure 1 that needs quality improvement.
In experiments section, test data is not clear. Only authors focused on training (70%) and remaining as validation data. This needs justification.
The presented findings are giving an impression of self-comparative results as in Figures. Therefore, question rises that the novel hybrid ( LSTM and RL) model should be compared with existing works for findings validity.
Finally the paper is concluded well.
Authors have focused on an interesting research field by combining LSTM and RL to propose a novel hybrid model that achieved 0.94 R-square average value.
This achievement requires comparison with recent existing works.
The paper is concluded well in a proper format.
All text and materials provided via this peer-review history page are made available under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.