OneTick Cloud

PERFORMANCE

Built for high-performance tick analytics at scale

• Multi-threaded 64-bit server architecture
• No limitations on data volumes, peak rates, or length of stored history.
• Able to handle global market data (over 10+ billion messages per day)
• Bulk processing rate of more than 10 million ticks per second, per core
• Real-time collection of 400K-600K messages per second, per core (depending on the data feed)
• Intraday loaders can insert about 600K messages per second per core (depends on the normalization complexity)

ADVANCED ANALYTICS

Engineered for today’s market data complexity

OneTick is purpose-built to handle the structure and nuance of trades, quotes, and full depth order books.

You get a broad library of built-in analytics components — called Event Processors — designed specifically for trading and quantitative research on tick data.

• Research the relationship between trades and quotes
• Merge order books from multiple venues in real time
• Convert currencies for cross-border arbitrage
• Apply hundreds of market-specific analytics functions out of the box

These Event Processors extend beyond core market data. You can apply them to executions, request-for-quote (RFQ) systems, and your own custom data structures to build analytics that reflect how you actually trade.

REAL-TIME ALERTS

Monitor execution quality and act in real time

Configure applications such as a real-time execution quality monitor to trigger alerts based on the conditions that matter to you.

Route those events into a configurable alert management dashboard where you can review, assign, comment on, and close execution quality alerts within a structured workflow.

SCALABILITY

Run real-time trade analytics at scale in any market condition

You can process and analyze tick data in real time, even during peak volatility and extreme volume events.

• Multiple physical tick servers each running 1+ tick server processes
• Virtual databases and multiple real-time collector mounts
• Any number of collectors on any number of nodes
• Any amount of storage accessible from the query engines (aka tick servers). Includes support for multiple file systems.
• Multiple multi-threaded tick server processes on one physical server

GRAPHICAL QUERY DESIGNER AND PYTHON SCRIPTING

Design complex queries with drag-and-drop logic

Use the Query Designer to construct and model queries graphically with reusable Event Processors. Event Processors act as nodes in a directed graph. You link them in series or parallel to build complex logic using an intuitive drag-and-drop canvas. Each processor includes its own configurable parameter set.

You can define parameters as static values, reference tick stream fields, conditional expressions, or runtime inputs such as aggregation intervals. For programmatic workflows, you can use our Python interface to create and run OneTick queries with the flexibility of Python and the performance of OneTick.

• Create and execute queries in Python
• Headless approach to development
• Support for commonly used packages (i.e. Matplotlib, NumPy, Pandas, and Jupyter notebooks)
• Directly access OneTick’s Cloud and /or locally stored data

MACHINE LEARNING

Run machine learning directly on tick data

You can use OneTick’s machine learning event processors for training, prediction, performance evaluation, and real-time learning/prediction on-the-fly. These operations can be performed natively within OneTick, without having to export any data.

• OneTick’s ML algorithms address common machine learning categories such as classification, regression, and clustering.
• Specific algorithms include support vector machines, relevance vector machines, kernel ridge regression, and k-means clustering.
• The trained models are themselves stored as OneTick databases.