Skip to main content
Back to challenges
Challenge

Deploy Real-Time Trade Surveillance for Regional Stock Exchanges with Semantic Kernel

Emerging regional stock exchanges require automated order-book manipulation detection. Develop a Semantic Kernel agent pipeline that monitors high-frequency equity trading for spoofing, layering, and wash trading anomalies in real time.

CybersecurityHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Build a real-time market surveillance agent using Microsoft Semantic Kernel to detect market manipulation patterns in equity order streams.

Delivery guide

How work is evaluated

Evaluation

Measures market manipulation detection precision on order book datasets.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

Spoofing Flag Test

Rapid large order submission followed by sub-50ms cancellation must flag SPOOFING.

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

Surveillance Precision

Precision of flagged market abuse events. • target: 0.97 • range: 0.9-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Implement a Semantic Kernel pipeline in Python or C# with custom native plugins for order book telemetry.

  • Detect manipulative order book behaviors including spoofing, layering, and wash trading.

  • Generate real-time FINRA/SEC-compliant market alert records.

  • Maintain low-latency throughput for streaming exchange order events.

Resources and assets

Reference links and supporting material

Dataset notes

Synthetic L2 order book data feed with injected market manipulation patterns.

Challenge setup

Measures market manipulation detection precision on order book datasets.

Challenge input

JSON list of order updates (submit, cancel, execute) with timestamp and trader ID

Semantic Kernel

Integrates native telemetry plugins with cognitive LLM reasoning.

Evaluated output

JSON with anomaly_detected (boolean), anomaly_type ('SPOOFING', 'WASH_SALE', 'NONE'), confidence

Evaluation checks
  • Rapid large order submission followed by sub-50ms cancellation must flag SPOOFING.
  • Precision of flagged market abuse events. • target: 0.97 • range: 0.9-1
Configured evaluation evidence
  • Benchmark: TradeSurveillanceBench
  • Surveillance Precision target: 0.97
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • Semantic Kernel · Required

Evaluation contract

  • Spoofing Flag Test · Weight 1
  • Surveillance Precision · Weight 1

Metrics

  • Surveillance Precision · Target 0.97

Benchmarks

  • TradeSurveillanceBench
  • security

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON list of order updates (submit, cancel, execute) with timestamp and trader ID

    Output: JSON with anomaly_detected (boolean), anomaly_type ('SPOOFING', 'WASH_SALE', 'NONE'), confidence

    Execution harness: Python

Evaluation tools

  • Semantic Kernel · Agent Framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Deploy Real-Time Trade Surveillance for Regional Stock Exchanges with Semantic Kernel