May 2018 Summaries

MarketStore is an open-source financial timeseries database used for storing and querying historical price data. The current setup uses a background worker to fetch minute-level historical prices, which provides faster data speed compared to the previous setup that fetched daily bar data. This increased data speed enables more backtesting and training in machine learning for trading algorithms. MarketStore offers scalability advantages over PostgreSQL, allowing it to serve multiple symbols and timeframes efficiently. It also reduces the need for complex queries, making it easier to query larger datasets. The performance benefits of using MarketStore were demonstrated through benchmarking tests comparing its speed with PostgreSQL.

Algorithmic trading, like software development, benefits from continuous testing against historical data to ensure performance. However, collecting and processing large datasets can be a challenge. MarketStore, a platform for storing and querying financial data, offers a solution by providing a Docker-based setup that allows users to test their trading strategies quickly and efficiently. The system consists of two containers: one for the server and another for the client, which uses Python and pymarketstore to query the data. By following this tutorial, users can set up MarketStore to fetch historical Bitcoin price data from GDAX and query it using JSON-RPC, enabling them to develop and test their trading algorithms with a large dataset.