Software: Order-flow and long-memory in a simulated financial market

<p dir="ltr">This repository contains the R scripts for our<a href="http://10.25375/uct.30453044" rel="noreferrer" target="_blank"> UCT Statistics Honours Project</a> investigating long-range memory and order-flow dynamics in high-frequency JSE Top40 data. The project's central goal is to empirically validate the Lillo, Mike, and Farmer (LMF) hypothesis, which posits a specific relationship ($\gamma \approx \alpha - 1$) between the exponent of the trade-sign autocorrelation ($\gamma$) and the exponent of the metaorder size distribution ($\alpha$). The code includes functions for cleaning and preparing <a href="http://10.25375/uct.30449312" rel="noreferrer" target="_blank">raw JSE trade data</a>, replicating theoretical LMF simulation models, and running the main validation analysis. Key scripts apply custom metaorder generation algorithms to the empirical data to estimate and compare the $\alpha$ and $\gamma$ exponents.</p>