A Data Mining Approach for the Analysis of 'Stock-Touting' Spam Emails
Mohamed Mahmoud Zaki, Babis Theodoulidis, David Díaz · Research Explorer (The University of Manchester) · 2010
Although the financial markets are regulated by robust systems and rules that control their efficiency and try to protect investors from various manipulation schemes, markets still suffer from many attempts to mislead or misinform investors in order to gain illegal profits. The impetus to effectively and systematically address such schemes is presenting many challenges to academia, industry and relevant authorities. This paper proposes the use of data mining techniques to detect abuses in the stock market and in particular, Information-based manipulations. These are manipulations that allow rogue traders to gain illegal profits from disseminating false or vague information to investors through spam emails. The paper proposes a spam fraud analysis and detection framework using data mining techniques that helps analysts to identify possible touting cases based on spam emails. The framework employs different techniques such as classification, neural networks and linear regressions. The application of the framework is demonstrated using data from the Pink Sheets market and the results strongly suggest that data mining techniques can be used to facilitate fraud investigations originating from spam emails.The proposed framework and findings of the paper could be used in a retroactive mode to help the relevant authorities and organisations to identify abnormal behaviors in the stock market. It could also be used in a proactive mode to warn analysts and stockbrokers of possible cases of market abuse.