Winner Strategies in a Simulated Stock Market

Round 1

Reviewer 1 Report

Comments on the manuscript submitted to the International Journal of Financial Studies

Title: “Winner Strategies in a Simulated Stock Market”

Originality:

It is worth investigating in this subject, and the authors attempted to add new insights into the field.

Strengths:

1. In general good study and good quantitative method to approach research question.

2. Authors made considerable effort in the empirical part of this the paper.

3. The authors picked an interesting topic relative to agent-based simulation in stock market

4. The authors did a good job, but literature review should be up to date.

Weaknesses:

1. Introduction is too short and do not well present the objective of the manuscript. The contribution of the authors must be clear and well written in the introduction. References presented in the introduction are very old (1934 and 1936). New references in the field should be included in the introduction. The rest of the paper should be presented briefly in the last paragraph of the introduction.

2. Literature review is well presented but I think it should be updated and authors should add new references in the domain.

3. Conclusion and policy implications:

Recommended to reiterate the main objective of the study. Little re-structuring of this section recommended to make it more reader friendly and stating clear and straight forward conclusions.

4. Generally, the whole study needs proper proof-read with regard to used language and grammar. Throughout the study it is not straight-forward to understand what is meant.

6. Some upgrades needed prior to publishing.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper addresses an interesting issue by asking what sort of investment strategies would do best in a multi-agent model where the agents are mostly rational but with specific investment style preferences.

The formulation is in part attractive because it allows expected returns to be interpreted differently by investors (representing different investment styles) along three dimensions with random ranges of parameters leading to different agents. The dimensions are duration for assessing a trend, trend following preference, and assumed speed of price convergence to fundamental.

The model makes a bunch of additional assumptions that are not quite standard and inconsistent with optimal agent choice.  What is called a "utility" specification is just considering the expected log wealth and does not consider utility from consumption. The assumption that half the investors have negative speed of convergence parameters is hard to understand. Most odd is the determination of V in eq. 5  This sets the fundamental value with a market equilibrium equation that really should be setting the price P.  A further assumption is that P=V which seems to throw out the baby with the bath water. It also implies an infinite speed of conversion counter to the assumptions made by the agents. Further, rather than making wealth or the portfolio return provide the score, the authors use an ad-hoc point system.  Also not clear why short-selling and borrowing are not allowed. Agents have an incentive to have zero probability of bankruptcy. Any "mistakes" allowing possible bankruptcy are the authors' modeling or programming errors.

The main result, that the best investment approaches are these with a low assumed speed of convergence and the second-best approaches these with a low speed of divergence, depends crucially on the odd assumptions made in the model set-up.  A better discussion of what we really learn from this is important as well.

The writing should be improved. First, the authors should use the common convention to use present tense rather than past tense. They forgot to state the distribution of the speed of convergence parameter (cat), and skipped other relevant details that would allow the study to be replicated. The intended contribution relative to existing literature and its importance were not properly presented in the introduction.  What is "indigenous price"? You need to state explicitly that the risk-free asset has zero return. This only becomes guessable when you suddenly introduce cash. No reason to include Table 1.  It is very unclear and you can easily explain what you are up to without this table.  Replace "trend following desire" by trend following preference.

 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The authors explained their approach better in this version and the paper is improved. However, I still am unclear about two very important issues.

1. The key issue is how the actual price P is determined. There is no equation for this in the paper but the simulation somehow must infer the price by setting total demand equal to a fixed supply. Of course this is needed to determine the realized return and then the investor score in the scoring scheme. 

2. In this approach I do not understand why a "fundamentally neutral market" is necessary: if all the cat values are positive and lead to demand either being too high or too low, then the price will adjust.

Minor:

a. fairness [bottom of p.4]. Explain why this is relevant. seems that this may just be a simplifying assumption.

b. middle of p.3 you state "hens (2018)"  this appears to be a reference that is not listed.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 3

Reviewer 2 Report

All my comments have been addressed.