Suspicious Banking Transaction Detection Using Shewhart X-Bar Control Chart

Financial organisation (i.e. bank) is constantly looking at method improving fraud detection to detect and stop the highest-risk transactions. This paper presents a Statistical Quality Control (SQC) approach, i.e. Shewhart X control chart for suspicious banking transaction detection based on bank indicator rules and control chart rules. One millionbank transaction records performed by 2000 accounts have been simulated. There are a total of 150 accounts randomly selected with 50 of them being indicated as suspicious transactions accounts. Two types of transaction records (i.e.deposit and withdrawal) were plotted on the Shewhart X chart in pair. A banking transaction is marked as suspicious when the chart indicates a matching pattern with bank indicator rule or the control chart rule. There are 45 out of the 50 accounts correctly detected as suspicious with 90% of sensitivity and 100% of specificity. Overall, the new method effectively detects suspicious transactions and improve client risk profiling.

Keywords:

suspicious transaction , SQC , control chart , sensitivity , specificity