Title: The accuracy of trade classification algorithms: Evidence from the Budapest, Prague and Warsaw stock exchanges
Authors: Sabina Nowak - University of Gdansk (Poland) [presenting]
Abstract: The aim is to evaluate and compare the accuracy of the transaction classification rules, including tick, reverse tick, quote, Lee and Ready and Ellis, Michaely and O'Hara rules, on the three Central and Eastern European stock exchanges: the Warsaw Stock Exchange, the Prague Stock Exchange and the Budapest Stock Exchange. The transaction data of 50 companies, the constituents of the blue chip indices, namely WIG20, PX and BUX, from October 2018 to June 2019, are employed. Tick data is collected from the Refinitiv (previously Thomson Reuters) database. The problem of the classification of transactions initiated by sellers and buyers as well as its accuracy is widely addressed in a current literature on the financial market microstructure. Most of the research concerns the US financial markets; research on other markets is scarce and usually includes a single country, as Australia, Brazil, Germany, Taiwan or Turkey. There is a lack of similar research devoted to the European developing markets. The aim is to fill this cognitive gap with reference to the selected markets of the Central and Eastern Europe.