Forecasting Agriculture Commodity Futures Prices with Convolutional Neural Networks with Application to Wheat Futures

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article appears to have been hastily assembled with little attention to detail. Further comments are provided in the attached document.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

Poor - extensive proofreading is required.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The authors analyse the use of convolutional neural network with wind images over wheat planting areas. The objective is to take short/long positions in the wheat futures market, based on 20-day forecasts.

- From a formal point of view, the article is well written. However, it would be advisable for the authors to indicate the source in both tables and figures. In addition, it would be advisable to review the incorporation of Doi in all the references that they do have. For example, reference 22. Kuan et al (1195) has doi https://doi.org/10.1002/jae.3950100403 and should be incorporated.

- In the literature review section, other studies that attempt to predict prices directly with neural networks whose inputs relate exclusively to prices can be incorporated. For example, García et al. (2018) present a hybrid fuzzy neural network to forcast price direction in a stock index. Qiu et al. (2020) use a recurrent neural network to predict different stock market indices. On the other hand, Yu et al. (2020) use a deep neural network with the time series phase-space reconstruction (PSR) method to predict multiple stock market indices.

 

 García, F., Guijarro, F., Oliver, J., & Tamošiūnienė, R. (2018). Hybrid fuzzy neural network to predict price direction in the German DAX-30 index. Technological and Economic Development of Economy, 24(6), 2161-2178. https://doi.org/10.3846/tede.2018.6394

Qiu J, Wang B, Zhou C (2020) Forecasting stock prices with long-short term memory neural network based on attention mechanism. PLoS ONE 15(1): e0227222. https://doi.org/10.1371/journal.pone.0227222

Yu, P., Yan, X. Stock price prediction based on deep neural networks. Neural Comput & Applic 32, 1609–1628 (2020). https://doi.org/10.1007/s00521-019-04212-x

 - In section 3, the authors indicate that images have been taken for the different geographical areas. It would be useful to explain whether this is one image per day for several days in order to have daily average climatological data, given that they indicate that daily futures price data have been used.

- It would be useful if the authors could justify, if possible, the 20-day price prediction objective and not another period.

- The sample period is from 1984 to 2014. It would be useful if it is possible to access the most current data.

 

Author Response

Thank you for the comments and suggestions. Please see the attached document for our response. 

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

The background has not been presented as to the fundamental issues of gap studies and research regarding the CNN method used

·      The literature review needs to be presented in a more updated manner to demonstrate this study's novelty and contribution. The CNN method has been widely applied in various empirical studies in the last five years.

·       The research results are not clearly presented, so they do not show good scientific fairness. A discussion of the results can be added to a separate sub-section that explains the model developed more deeply and broadly. Besides that, the applied model must show the position and contribution of existing and tested empirical models.

·       The abstract and conclusion are written the same; they should be presented differently. Conclusions include study limitations and suggestions for a future research agenda.

·       The paper can add references to the last five years to strengthen the literature review and discussion of research results.

Author Response

Thank you for your kind comments and suggestions. Please see the attached document for our replies. 

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Considerably improved - now reads like a very interesting paper

Author Response

Thank you again for all the comments and suggestions that greatly improved this paper. 

Reviewer 3 Report

Comments and Suggestions for Authors

• The paper has revealed two contributions of the study: using image data as a fundamental proxy in estimating agricultural commodity prices and direct trading. Contribution statements should begin with criticism of previous work that needs to be refined.

• Updated literature reviews increasingly show the novelty and contribution of research.

• The research results have been presented clearly but need to be provided with in-depth discussion to strengthen the argument of the empirical findings.

• The conclusion includes essential points from the research results and is different from the abstract in terms of writing.

Author Response

Thank you for your suggestions. We've added more details in the introduction section (lines 24 - 27, 53 - 59, 63 - 67), and the discussion of results (lines 309 - 311, 359 - 367). In particular, we highlighted the implications of our findings for commodities that are highly financialized, and/or thinly traded as they're more likely to be influenced by crowding trades. 

Again, we are grateful for all your comments and suggestions. They've helped us improve this paper drastically.