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.
The reviewers consider their comments as properly addressed therefore your manuscript can be accepted for publication.
no comment
no comment
no comment
Thank you for taking so many of my improvement suggestions into account.
I have no further issues with the manuscript.
Good Work!
No comment
no comment
no comment
The authors have provided a clear and detailed explanation of all the points raised. For this reason I believe that the article is now ready for publication.
After considering the reviews made by the reviewers, there are some aspects regarding the methodological aspects and presentation of results that need to be addressed. Please, resubmit the paper after incorporating the above-mentioned improvements.
[# PeerJ Staff Note: Please ensure that all review comments are addressed in a rebuttal letter and any edits or clarifications mentioned in the letter are also inserted into the revised manuscript where appropriate. It is a common mistake to address reviewer questions in the rebuttal letter but not in the revised manuscript. If a reviewer raised a question then your readers will probably have the same question so you should ensure that the manuscript can stand alone without the rebuttal letter. Directions on how to prepare a rebuttal letter can be found at: https://peerj.com/benefits/academic-rebuttal-letters/ #]
no comment
A weakness of the manuscript is the unclear objective of the paper. 5 research questions are presented without explanation why these questions are relevant. So it is not clear if the paper is about to find the best algorithm for a specific asset (BTC) or to compare profitability of specific assets using the algorithms.
Please do more explanation why you use this 5 research quesitions. And make the objective of the paper clearer.
no comment
1. The algorithms 1 – 4 generate buy and sell signals even if it not possible to buy or sell the assets. Please change your algorithms in a way that they consider previous prices and only generate buy or sell signals, if it is possible to buy respectively sell. You cannot sell without buying before. Specially for algorithm 4 you even describe, that buy and sell signals are generated if two moving averages cut each other (lines 192 – 195). Nevertheless, in algorithm 4, signals are generated, if one moving average is higher/lower than the other.
2. In line 206 – 208 you describe that algorithm 4 generate signals without a cut of the two moving averages. Please be consistent.
3. In this context, table 2 should be adjusted. Here you show the number of signals which are generated over the whole data set per algorithm. There is a large number of buy and sell signals but the number of completed transactions is very small. It seems you assume that if the last day of a trading period occurs, it is not necessary to sell all units bought before. This is a typical assumption in the literature of competitive algorithms which you refer to. Note, if there is no need to sell the units which are bought before, your benchmark (the buy and hold strategy) is in disadvantage. Specially, if transactions costs are considered.
4. Further it is not clear, if you take the return of trading periods into account in which only buy signals are generated.
5. If all bought units have to be sold at the last day of a trading period, this should also be noted in algorithms 1 – 4.
6. Why is FLMA the only algorithm which is not presented?
7. Another important issue with your manuscript is the relevance of your research questions. The five research questions are listed in the lines 141 – 145 without any explanation why these questions are relevant.
8. One unanswered question is how do you determine m and M for RP?
9. Figure 2 shows the transactions per algorithm. Figure 1 shows the effect of transactions costs. Because the results of figure 1 depend on the findings of figure 2, I would change the order of figures and show figure 2 first. Further, this change in order gives you the opportunity to discuss the findings of figure 1. This could be very helpful, because at the first sight it seems, that the algorithms with a low number of transactions, haven’t a lower impact from transaction costs.
10. Table 2 is a summary of the selected scenarios but not of the algorithms. In fact, you use six algorithms, in two different settings (15 days, 30 days). Further, it is not clear, why you choose different parameters for the algorithms VLMA and FLMA for 15 days and 30 days. Why do you not use the four combinations of parameter for 15 and 30 days?
11. In line 203 you define e_t again, without using it after that.
12. Please define the first two variables in line 209 before using it.
13. Please define the variables i and f before using it. Line 226 -228. Why are you not using T?
14. Please define the variable T before using it. (First use in algorithm 2)
15. Please define the variable q_t before using it. (First use in algorithm 2). Do you mixed up q_t and e_t?
no comment
no comment
no comment
The authors investigated the short term profitability of Bitcoin cryptocurrency against two fiat currencies, namely Euro and Yen.
The analysis was conducted implementing both reservation price algorithms and moving average based trading techniques over an eight year period, ranging from January 1, 2011 to 31 December, 2018.
The results were compared with those obtained with the classical buy-and-hold trading technique and were evaluated in terms of the (Geometric) average trading period return criterion.
This study highlights that on average, in short term Yen is more profitable than Bitcoin and Euro, although the answer also depends on the choice of the algorithm.
The article is clearly written and the analyses seem to be performed rigorously. Raw data and code are provided by the authors and easily accessible. For these reasons, I think that the paper would be of interest for the readership of PeerJ Computer Science.
However, there are some minor changes that must be addressed for the article to be ready for publication.
The "Introduction" of the paper should be proofread. I suggest to move the content of section 2 (Bitcoin - A brief overview) to the "Introduction" section, describing in a more detailed way the Blockchain technology, focusing more on its main feature of public ledger. The authors made a comparison between Bitcoin cryptocurrency and two fiat currencies, namely Euro and Yen. I think that they didn't highlight the main problem of this comparison, which is also the main feature that distinguish their markets, i.e. fiat currencies depends only on classical macroeconomic variables while cryptocurrencies are virtual currencies based on Blockchain technology and therefore their market also depends on variables related to the technology itself. Please go into this in more detail in the "Introduction" section.
Figures 1 and 2 shows the results obtained for the Bitcoin cryptocurrency. If there are no length limits, why not also report the figures for Euros and Yen? Please, report all results.
The purpose of this paper is to ascertain the short term profitability of Bitcoin comparing it with those of the Euro and Yen fiat currencies. This comparison gives a sense of how different BTC is. Actually, to make this analysis more robust I would include the analysis of another cryptocurrency in order to have an equal comparison, for example (Bitcoin, Ethereum) VS (Euro, Yen). The choice of BTC and ETH is quite straightforward, since they are the two most valuable cryptocurrency at the moment. Please, expand the experiment with one more cryptocurrency.
- Page 4, table 1. I suppose the data are in USD, but please always report the unit of measure or specify it in the table description.
- Page 7, section 5.2. The authors should be more accurate in the description of the chosen evaluation criterion. Equation 6 doesn't represent the Average Trading Period Return, which is the simple arithmetic mean of returns, but instead the Geometric Average Trading Period Return, that is the geometric mean of returns. Furthermore, the authors should explain how this evaluation criterion should be use in this study. For example, when dealing with time series prediction we use MAE (Mean Absolute Error) error to evaluate results. A lower MAE means better performance, while a worse MAE means worse results. Please, clarify these points.
- Page 7, lines 221-224, line 235. Page 9, line 285. The authors should explain how and why they choose this values. Did the authors use or follow a particular criterion for the choice of these values? For example, when dealing with clustering the number of cluster to build is usually chosen because of the Elbow method. This is a critical issue that the authors should clarify.
- Page 11, Figure 2. Please, choose "Signal" or "Signals".
- Page 11, line 330. "whereas moving based algorithms...". Is "whereas moving average based algorithms..." what the authors would like to say?
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.