Discussion Paper
No. 2017-63 | September 25, 2017
Joeri Schasfoort and Christopher Stockermans
Fundamentals unknown: momentum, mean-reversion and price-to-earnings trading in an artificial stock market
(Published in Agent-based modelling and complexity economics)

Abstract

The use of fundamentalist traders in the stock market models is problematic since fundamental values in the real world are unknown. Yet, in the literature to date, fundamentalists are often required to replicate key stylized facts. The authors present an agent-based model of the stock market in which the fundamental value of the asset is unknown. They start with a zero intelligence stock market model with a limit-order-book. Then, the authors add technical traders which switch between a simple momentum and mean reversion strategy depending on its relative profitability. Technical traders use the price to earnings ratio as a proxy for fundamentals. If price to earnings are either too high or too low, they sell or buy, respectively.

JEL Classification:

C63, D53, D84, G12, G17

Cite As

Joeri Schasfoort and Christopher Stockermans (2017). Fundamentals unknown: momentum, mean-reversion and price-to-earnings trading in an artificial stock market. Economics Discussion Papers, No 2017-63, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2017-63


Comments and Questions



Anonymous - Referee Report 1
October 19, 2017 - 08:05
see attached file
[ Download attached file ]

Joeri Schasfoort - Reply to referee
November 03, 2017 - 11:04
We thank the anonymous referee for these constructive comments. We would like to incorporate all of them in the next version of the paper. In the following section, for every remark, we will elaborate how. 1) We recognise that the calibration procedure is relatively primitive. Currently, we run different simulations over a sample of the full parameter space. We then minimise the difference for the conditions of: no autocorrelation, fat tails, volatility clustering, and long-memory. To obtain a better fit for Kurtosis, Autocorrelation abs and the Hurst index, we propose to use an optimization technique. Following Thiele et al. (2014), for a revised version of the paper, we propose to calibrate the model using a genetic algorithm. Similar to what we did before, we will first sample the parameter space using a Latin Hypercube and determine their fitness by comparing the simulated stylized facts to actual stylized facts. Then, we will cross breed the fittest parameter settings with some random other parameter sets (to avoid getting stuck in a local maximum). This process will deliver a fitter sample of parameter combinations. We will repeat this process until it stops improving, when it finds the global optimum fit. Thiele, Jan C., Kurth, Winfried and Grimm, Volker (2014) 'Facilitating Parameter Estimation and Sensitivity Analysis of Agent-Based Models: A Cookbook Using NetLogo and 'R'' Journal of Artificial Societies and Social Simulation 17 (3) 11 . doi: 10.18564/jasss.2503 2) We believe our paper could benefit from a comparison to other models (a practice also encouraged by Fagiolo, Moneta & Windrum 2007). We read the suggested papers. We will compare our results to those of the Brock Hommes Model, which was also used in the suggested papers. Fagiolo, Giorgio, Alessio Moneta, and Paul Windrum. "A critical guide to empirical validation of agent-based models in economics: Methodologies, procedures, and open problems." Computational Economics 30.3 (2007): 195-226. 3) We will add these to a revised version. 4) This never happens in our simulation. If the price is falling toward the lower P/E ratio bound, demand increases and the price will rise. Yet, a scenario in which this does not happen is not impossible. This is thus a possible source for bugs. We will update the price decision rule so that this is no longer possible. 5) The equation should indicate that order volume is equal to an agents money / price if the offer type is a ‘bid’ and that volume is equal to the total amount of stocks in the agents possession if the order type is ‘sell’. We will update the equation to accurately represent this. 6) We will add this. 7) We intended the discount rate to be annually. However, given that our profits are monthly. These discount rates are indeed not realistic. We will adjust them accordingly.

Anonymous - Referee Report 2
October 25, 2017 - 11:50
REVIEW REPORT Paper: Fundamentals unknown ... stock marketAuthors: J. Schasfoort & C. Stockermans The paper present an agent-based framework for analysing the stock market dynamics, using some proxy forfundamentalists and chartists for the decision making of the agents. Some encouraging simulation results are alsopresented. The study seems interesting and new. However, the framework employed here should be compared with that in a detailed study reported earlier (2014): Confidenceand the Stock Market: An Agent-Based Approach, M. A. Bertella, F. R. Pires, L. Feng, H. E. Stanley, PLoS ONE 9(1): e83488.doi:10.1371/journal.pone.0083488 The paper need to be revised comparing the approaches and highlighting the major distinguishing and promising featureshere.

Joeri Schasfoort - Reply to referee
November 03, 2017 - 11:24
We would like to thank the reviewer for pointing out this paper to us. We where not aware of its existence. It's addition of a behavioural bias in the form of confidence levels is especially interesting. We intend to update the paper with a comparison to other models, we will include this model in that comparison.