View a markdown version of this page

Algorithmic trading on AWS - Financial Services Reference Architecture Diagrams

Algorithmic trading on AWS

Publication date: December 21, 2020 (Diagram history)

With this architecture, you can backtest and host machine learning (ML)-based algorithmic trading strategies. The solution uses Amazon Elastic Container Service with AWS Fargate for containerized services, Amazon SageMaker AI for model training, and Amazon DynamoDB for transactional data storage.

Algorithmic trading on AWS diagram

Price data flows through Athena to train a model in SageMaker AI and run trading strategies on Amazon ECS with DynamoDB.

The following steps describe the data flow and trading components for this architecture:

  1. Get historical price data from AWS Data Exchange or external market data sources.

  2. Catalog historical price data in AWS Glue Data Catalog and query data by using Amazon Athena.

  3. Train the ML model for ML-based trading strategies by using SageMaker AI.

  4. Run the backtest for your trading strategy as a task through AWS Fargate and Amazon ECS.

  5. Run the Market Data Adapter as a service through AWS Fargate and Amazon ECS. Feed data into DynamoDB.

  6. Run the Broker Adapter as a service through AWS Fargate and Amazon ECS. Use DynamoDB as the transaction store.

  7. Run the Trading Strategy as a service through AWS Fargate and Amazon ECS.

  8. Use Amazon EventBridge for job scheduling with AWS Lambda functions. Use Amazon CloudWatch for monitoring.

Further reading

For additional information, see the following resources: