Dear Mahmut Baydas, Tevfik Eren, Željko Stević, Vitomir Starčevic

I read your paper

Proposal for an objective binary benchmarking framework that validates each other for comparing MCDM methods through data analytics

My comments

1- I am not sure that the title is correct since benchmark is defined by the Dictionary as “a standard or point of reference against which things may be compared or assessed”

If you compare two results, obtained analytically what is happening is that they are checking each other, but it does not imply that the coincident result is correct, regarding the real thing, although it speaks favourably of the method in comparison to others.

2- As I understand, ‘stock return’ is the “the positive or negative change in value of an investment or asset over time” (Money Sense Editors, February 27 2023).

As well, ‘financial performance’ is “A complete evaluation of a company's overall standing in categories such as assets, liabilities, equity, expenses, revenue, and overall profitability (CFI Team)”

These are criteria, evaluating 140 companies or alternatives. You want to demonstrate that both criteria validate each other, therefore, you are establishing that a correlation may quantify this relationship. In my opinion, the word ‘validate’ is incorrect. Correlation shows that both variables move in a large stent in the same sense in their ups and downs

, but this is not validation, because you are not comparing them with a known yardstick.

Then you use nine different MCDM methods, and what do you get out of that? Since all methods work with the same initial matrix, the existing relationship does not change, although what can change is the ranking of alternatives.

Therefore, what is the purpose of the article?

After that, you introduced the RR concept, which I don’t see its relationship with the former problem.

You say “ According to the findings obtained entirely through data analytics, Faire Un Choix Adéquat (FUCA) and (which is a fairly new method) the compromise ranking of alternatives from distance to ideal solution (CRADIS) were determined as the most appropriate methods by the joint agreement of both criteria”

Using FUCA and CRADIS may indicate what you say, but where is the proof?

3- In page 3 “Simply put, MCDM methods can be compared based on their ability to relate to real life.This brings to mind the naturally occurring sequences in real life”

This is a most important definition of MCDM methods.

4- I have only notions on financing, but in my understanding this sentence is incorrect “The share price is an example, and similar rankings can also be used in other scientific fields”

In MCDM we work with facts that derive from the scenarios, like calculated costs, estimated gains, wished ROE, estimated working capital, and all of them are related through the projected financial statements, etc., and are mostly constant, at least during a certain period, but the share price does not belong to this category, because it is highly dynamic, depends on the markets, social issues, wars, etc., on what we don’t have any control. I think I discussed this same issue with Mr. Baydas long time ago.

Page 3“MCDM inputs; a character analysis based on MCDM outputs can show us a clearer path. In this sense, “price”,

For ‘price’ in financing I understand that you refer to “the price for which the stock is purchased, and thus, becoming the new market price (Investopedia)”

In my opinion what you say is incorrect, for the price depends on thousands of people buying and selling. Of course, it is a part of real-life, but that we cannot control, like cost, working capital, expenses, etc., and it is not similar to performance, as you say, fundamentally because its intangibility (the purchaser is buying hope), and dynamicity.

Not surprising, you say the same, thus, how can you compare performance to price?

You say “Moreover, the level of this existing relationship varies according to the ability of MCDM methods”

Perhaps you should explain how.

Page 4 “In this study, an “output-based” solution obtained with “data analytics” is proposed as an alternative to a classical “input-based” methodological solution’

In this, I agree in a 100%

5- In page 5 you mention SWOT analysis on MCDM methods.

In general, what is made is the opposite, starting with a SWOT analysis that identifies different strategies, MCSDM selects the best of them

6- “For this reason, it has been suggested to apply more than one method to the same problem in order to give a more comprehensive result to the decision makers and it is still widely applied”

And what is it good for?

7- page 6- “On the other hand, compared to its competitors, AHP (Analytic Hierarchy Process) can produce consistency, time and energy problems as it makes many pairwise comparisons with subjective expert opinion”.

Energy problems are not based on opinions but in a set of conditions or criteria to be met like demand, cost, load factor, reliability, efficiency, equipment, atmospheric conditions, environment, health, new development, etc. Do you think that all of these criteria are subject opinions

8- RR is not due to normalization, but to passing from one dimensional space to another. That is, adding an alternative C to two alternatives A and B (in 2D,) transforms in in a different space (3D) and vice versa, according to my hypothesis

9- If we have to select a MCDM from the axiomatic point of view, probably none would be selected

The best MCDM method for a certain problem is which best can model all the characteristics of the problem.

10- In page 7 “In MCDM-based FP studies, there is generally a lack of objective justification for choosing and using an MCDM method”

Why? Can’t you establish a value for IRR, or NPV or Recovery time or whatever other financial index?

11- In page 8 you speak of measuring sensitivity to RR. It is za very interesting concept, and I will investigate it

12- In page 8 “It is a difficult and chronic problem to determine which of the MCDM methods is more suitable and which one should be chosen”

Again, it is no difficult. The only thing to do is to have all characteristics of the problem, and look for the MCDM that best adapt or can model them all. For instance, if you have exclusive alternatives, you must use binary performance values. Just look for the method that can handle them. For simple and personal problems, probably the best method is AHP , because it considers personal points of view, that not other method does.

These are my comments, hope they can help you

Nolberto Munier