Financial Anomaly Detection for the Canadian Market

Paper: arXiv 2604.02549

Authors: Luigi Caputi, Nicholas Meadows

Abstract

In this work we evaluate the performance of three classes of methods for detecting financial anomalies: topological data analysis (TDA), principal component analyis (PCA), and Neural Network-based approaches. We apply these methods to the TSX-60 data to identify major financial stress events in the Canadian stock market. We show how neural network-based methods (such as GlocalKD and One-Shot GIN(E)) and TDA methods achieve the strongest performance. The effectiveness of TDA in detecting financial anomalies suggests that global topological properties are meaningful in distinguishing financial stress events.

Complexity vs Empirical Score

• Math Complexity: 6.5/10
• Empirical Rigor: 6.0/10
• Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper presents a solid comparative study applying advanced mathematical methods like Topological Data Analysis (TDA) and Graph Neural Networks to market anomaly detection. While the focus is applied to the TSX-60, combining topological features with deep learning provides non-trivial insights into financial stress events.

Research Flowchart

  flowchart TD
    A[Goal: Detect Financial Anomalies in Canadian Market] --> B[Input Data: TSX-60 Equities Time Series]
    B --> C[Methodology: Evaluate 3 Model Classes]
    C --> D1[Topological Data Analysis - TDA]
    C --> D2[Principal Component Analysis - PCA]
    C --> D3[Neural Networks: GlocalKD & One-Shot GIN-E]
    D1 & D2 & D3 --> E[Detect Major Financial Stress Events]
    E --> F[Findings: TDA & Neural Networks Outperform PCA; Global Topology is Key]