How much do the central bank announcements matter on financial market? Application of the rule-based trading system approach. [Dataset]

Abstract

Trading systems constructed on financial markets, which are designed for prediction of financial instruments prices and investing in financial assets, can be classified according to the following two broad principles: the first group concerns entirely rule-based systems, while the second approach is more flexible and also more subjective, where only some general guidelines are used but there are typically no fixed rules adopted. In this paper, we propose a rule-based trading system related to the fuzzy logic reasoning in order to investigate how much the central bank's announcements matter on a financial market. The design of the prediction systems for financial markets and the analyses of performance of trading strategies, based on the fuzzy logic rules, neural networks, pattern recognition techniques or other artificial intelligence methodologies, have been subject of research in the existing literature, which expanded during the past two decades. We propose in this paper a rule-based trading strategy relying on the signals extracted from the NBP central bank's announcements regarding its monetary policy decisions. This approach is conceptually close to some fuzzy expert systems (see e.g. Dymova et al. (2012) or Rubell and Jessy (2016), among others) and the stock market investment strategies based on rule-based reasoning (see e.g. Brzeszczyński and Ibrahim (2019)). In this paper, we proposed an investment strategy based on the trading rules to investigate the performance of investments on a broad financial market in Poland in its main three segments in response to central bank's communication of its monetary policy decisions. The results of our analysis, relying on the application of the rule-based trading system, allowed to establish how much the central bank's announcements matter on a financial market. We designed a novel investment strategy and we simulated trades, which enabled us to quantify their profitability in the out–of–sample period using the data from a broad financial market in Poland spanning across 3 segments: stock market, foreign exchange market and bonds market. Our results show evidence that the individual transactions delivered profits in 72.7% cases. The overall profitability across all events and all trading horizons was positive in as many as 63.6% cases.

Citation

BRZESZCZYŃSKI, J., GAJDKA, J., SCHABEK, T. and KUTAN, A.M. 2021. How much do the central bank announcements matter on financial market? Application of the rule-based trading system approach. [Dataset]. Hosted on Mendeley [online]. Available from: https://doi.org/10.17632/3ysyh9vkwp.1

• Acceptance Date: May 11, 2021
• Online Publication Date: May 23, 2021
• Publication Date: Nov 15, 2021
• Deposit Date: Mar 31, 2023
• Publicly Available Date: Mar 31, 2023
• Publisher: Elsevier
• DOI: https://doi.org/10.17632/3ysyh9vkwp.1
• Keywords: Financial market; Fuzzy reasoning; Rule-based database; Rule-based trading system; Central bank; Monetary policy; Stock market; Foreign exchange market; Bonds market
• Public URL: https://rgu-repository.worktribe.com/output/1925322
• Related Public URLs: https://rgu-repository.worktribe.com/output/1893465 (Journal article)
• Type of Data: XLSX and accompanying TXT file.
• Collection Date: Feb 15, 2020
• Collection Method: The sources of data, which we exploited for construction of the database used in this study, are: National Bank of Poland (information about the NBP announcements dates and the values of the newly revealed data) and Bloomberg (data about prices from the foreign exchange market, bonds market and stock market). Our data sample covers over 10–years long period from 6th November 2009 to 15th February 2020. In order to test the proposed rule-based trading system in practice, we distinguish the in–sample estimation period, which ends on 24th May 2019 and which includes a total of daily 2491 observations, as well as the out–of–sample period, which spans from 25th May 2019 to 15th February 2020. The frequency of data is daily and we additionally included the intra-daily data (at 1-hour, 30-minutes and 1-minute frequencies) in the presented out–of–sample analyses. For the purpose of the design of our rule-based trading strategy, we first constructed regression models for 12 financial instruments from 3 most important market segments on the broader financial market in Poland: stock market (stock indices: WIG, WIG20 and sWIG80), foreign exchange market (currency exchange rates: USD/PLN, EUR/PLN, GBP/PLN, CHF/PLN and JPY/PLN) and bonds market (1–year bonds, 2–years bonds, 5–years bonds and 10–years bonds).