TRTH JSE AGLJ.J Intraday Transaction Test Data

Published: 02-05-2019| Version 2 | DOI: 10.17632/4rrk89c3b2.2
Contributors:
Timothy Gebbie,
Edward Nonyane

Description

An example of TRTH intraday top-of-book transaction data for a single Johannesburg Stock Exchange (JSE) listed equity. The data is for teaching, learning and research projects sourced from the legacy Tick History v1 SOAP API interface from https://tickhistory.thomsonreuters.com/TickHistory in May 2016. Related raw data and similar data-structures can now be accessed using Tick History v2 and the REST API https://hosted.datascopeapi.reuters.com/RestApi/v1. Configuration control: the test dataset contains 16 CSV files with names: "<Ticker>_<STARTDate>_TO_<ENDDate>.CSV" e.g. AGLJ_J_02-Jan-2014_TO_29-May-2014.csv Attributes: The data set is for the ticker: AGLJ.J from May 2010 until May 2016. The files include the following attributes: RIC, Local Date-Time, Event Type, Price at the Event, Volume at the Event, Best Bid Changes, Best Ask Changes, and Trade Event Sign: RIC, DateTimeL, Type, Price, Volume, L1 Bid, L1 Ask, Trade Sign. The Local Date-Time (DateTimeL) is a serial date number where 1 corresponds to Jan-1-0000, for example, 736333.382013 corresponds to 4-Jan-2016 09:10:05 (or 20160104T091005 in ISO 8601 format). The trade event sign (Trade Sign) indicates whether the transaction was buyer (or seller) initiated as +1 (-1) and was prepared using the method of Lee and Ready (2008). Disclaimer: The data is not up-to-date, is incomplete, it has been pre-processed; as such it is not fit for any other purpose than teaching and learning, and algorithm testing. For complete, up-to-date, and error-free data please use the Tick History v2 interface directly. Research Objectives: The data has been used to build empirical evidence in support of hierarchical causality and universality in financial markets by considering price impact on different time and averaging scales, feature selection on different scales as inputs into scale dependent machine learning applications, and for various aspects of agent-based model calibration and market ecology studies on different time and averaging scales. Acknowledgements to: Diane Wilcox, Dieter Hendricks, Michael Harvey, Fayyaaz Loonat, Michael Gant, Nicholas Murphy and Donovan Platt.

Files

Steps to reproduce

The test data stored in *.csv flat files prepared using the discontinued v1 Tick History SOAP API. The trade sign is based on the Lee-Ready method to determine buyer or seller initiated trades. For a more complete description of the top-of-book data-pre-processing for a selection of BRICs data-sets with the same methodology see: 1. E Nonyane, (2019), MSc dissertation, WITS, "Calculating the price response of stocks in emerging markets" , (for related MongoDB interfaces and examples see https://github.com/Telmakaza/mongoDB_Tickstore) For JSE order-book data pre-processing see: 2. D Hendricks, (2016), PhD thesis, WITS, "An online adaptive learning algorithm for optimal trade execution in high-frequency markets" (see http://hdl.handle.net/10539/21710) The related current raw data and data-structures can be accessed using Tick History v2 and the REST API https://hosted.datascopeapi.reuters.com/RestApi/v1. The authors have no competing interest in relation to the use of raw or processed Thomson Reuters Tick History (TRTH) data and do not endorse TRTH in anyway.