Building a complete simulation trading and risk management system from scratch — the kind that powers real trading desks.

This project demonstrates the full lifecycle: from order entry and trade matching to real-time risk calculation and dashboard delivery. It’s designed to mirror the systems used at real trading desks, simplified for demonstration but architecturally faithful.

Overview

The system has two main subsystems:

  1. Trading System — an order book engine that accepts orders, matches trades, and stores executions in a Time Series Database
  2. Risk System — a real-time risk engine that computes PnL, Greeks, and other risk metrics, pushing updates to dashboards via an event queue

Both systems support multiple asset classes: FX, bonds, equities, options, and derivatives.

Architecture

Trading & Risk Management System Architecture

Trading System

Order Book Engine

  • Limit order book with price-time priority matching
  • Supports market, limit, and stop orders
  • Multi-asset: each instrument maintains its own order book

Trade Matching

  • Continuous matching for equities and FX
  • Auction matching for derivatives
  • Partial fills and order amendments

Trade Storage

  • Time Series Database as the columnar store for high-throughput trade ingestion
  • Designed for time-series queries: trade history, blotter views, aggregation by desk/portfolio

Risk System

Risk Engine

  • Real-time computation triggered by trade events
  • Calculates per-trade and per-portfolio risk metrics
  • Supports different calculation profiles per trading desk

Risk Metrics

  • PnL — Mark-to-market, realized vs unrealized
  • DV01 — Dollar value of a basis point (interest rate sensitivity)
  • Greeks — Delta, gamma, theta, vega, rho for options
  • VaR — Value at Risk for portfolio-level risk monitoring

Limits Framework

  • Per-desk risk limits (e.g., max notional, max DV01)
  • Per-portfolio limits
  • Breach alerting and pre-trade limit checks

Event Pipeline

  • Trade booked → event published to queue
  • Risk engine consumes event, calculates risk
  • Risk numbers stored in Time Series Database
  • Dashboard receives notification to refresh

Multi-Asset Coverage

Asset Class Instruments Key Risk Metrics
FX Spot, forwards, swaps PnL, delta, DV01
Bonds Government, corporate PnL, DV01, duration, convexity
Equities Stocks, ETFs PnL, delta, beta
Options Vanilla calls/puts PnL, delta, gamma, theta, vega
Derivatives Futures, swaps PnL, DV01, Greeks

Each trading desk sees only the risk metrics relevant to their book. An FX desk sees DV01 and spot delta. An options desk sees the full Greeks panel. The dashboard is configurable per desk.

Technology Stack

Component Technology Why
Core Engine Java Low-latency, strong concurrency
Message Queue MessageQueue High-throughput event streaming
Time-Series DB Time Series Database Columnar storage, fast aggregation
Caching CachingService In-memory grid for hot risk data
Dashboard React Real-time UI with WebSocket updates

Trading Desk Dashboards

Different desks have different views:

FX Desk

  • Live FX rates and spot positions
  • DV01 exposure by currency pair
  • PnL by book, intraday and cumulative

Rates Desk

  • Bond portfolio duration and convexity
  • DV01 ladder by tenor bucket
  • Yield curve shifts and PnL attribution

Options Desk

  • Full Greeks dashboard: delta, gamma, theta, vega
  • Volatility surface and skew
  • PnL decomposition by Greek

What You’ll Learn

This project demonstrates:

  • How order books work and how trades are matched
  • Real-time risk calculation patterns used in investment banks
  • Time Series Database schema design for financial data
  • Event-driven architecture for low-latency risk systems
  • How trading desks consume risk data differently

Repository

The full source code is available on GitHub: diepdaocs/trading-risk-system

Updated:

Comments