Proj S13 – FinSurge – FinSurge AI

Introducing Proj S13 – FinSurge – FinSurge AI

Team members

Joshua John Lee Shi Kai (ISTD), Cheyenne Chua Yan Ting (EPD), Khoo Zi Qi (ISTD), Li Xing (ESD), Sean Ho Junxian (ISTD), Aishwarya Shivakumar Iyer (ISTD)

Instructors:

  • Perry Wee Hian Lam

Writing Instructors:

  • Belinda Seet

FINSURGE AI TRADE SURVEILLANCE PLATFORM. DETECT FRAUDULENT TRANSACTIONS IN THE MARKET WITH THE POWER OF AI.

Fraudulent trades are financial transactions executed with intent to deceive, manipulate, or gain unfair advantage in a financial market.

Common characteristics include

  • Abnormal price spikes
  • Sudden volume surges
  • Repeated trading between related accounts

Fraudulent trade is HARMFUL

  • Reduces market integrity
  • Harm retail investors
  • Create distrust in financial system

Problem Statement

How might we design a web-based platform to strengthen fraud detection capabilities for financial institutions, compliance officers and credit analysts to detect suspicious transactions, activities, and prevent financial crime?

Solution

Web Platform

Our solution is a single web platform fraud detector that takes a trade dataset, applies models that classify each trade as normal or suspicious, then aggregates the flagged events into a human‑readable report.

Data → ML Models → Alerts

Types of Frauds:

  • Position Parking: Temporarily transferring securities to a counterparty to hide true ownership, then reversing the trade.
  • Window Dressing: Buying high-performing assets at period-end to make a portfolio look better than it was.
  • Off-Market Rate: Executing trades at prices materially away from market to transfer value between parties.
  • Marking the Close: Trading aggressively in the final minutes to artificially influence the settlement or closing price.
  • Floor/Ceiling Creation: Placing orders to artificially pin prices at a desired level, preventing natural market movement.
  • Ramping: Executing a series of trades to artificially drive a price up or down to benefit a related position.

How to use the Web Platform?

  1. Select the Trade Dataset
  2. Select the type of fraud
  3. Click on Run Detection
  4.  Detection Results, Model Visualisation, and Analysis Report will be produced, and the option to download the results in a CSV format, or format is available
  1. Click Download Report, and a report will be produced, so that financial institutions, compliance officers and credit analysts could comprehend
  2. An email will be notified to them to update them on the Trade Surveillance Report 

Capstone Video

Menu

ornament-menu