Algorithmic Trading: Custom Software Development

Algorithmic Trading

DataArt's services have been very helpful to firms and institutions which use automated trading strategies and quantitative data analysis. We help clients in building and implementation of alpha, risk, transaction cost and portfolio models.

For many modern strategies which use data mining for building data driven alpha models, intraday and end-of-day data availability and quality is critical for strategy success. High volumes of trades could lead to delays of historical data series delivered by exchanges. This often leads to losing days of algo trading.

DataArt is familiar with building custom frameworks and systems for big volume high frequency data logging:

  • High frequency market data signals (frequency down to five milliseconds)
  • Process up to 15,000 inputs (prices, Volumes and Quotes) for 5000 of US stocks per registration node
  • Real-time data filtering
  • Publishing filtered signals to enterprise message bus

Quick access to historical data series is another critical point for quantitative data mining. The speed of access to data often defines how much data one can mine and quality of output results: the more in-sample data is mined the better the out-of-sample strategy behavior will be. DataArt is experienced in building historical tick data storing in an optimized/compact format:

  • One-second data available for a three years period, thousands of signals, regular disk space
  • Quick access to historical data (access time is 600 ms when accessing one year of data for one signal)
  • Ability to apply cleaning (anti-spike) filters and transformation to raw historical data on the fly
  • Data quality validation jobs (missing points, duplicates, out-of-range values)
  • Integration jobs

DataArt also helps with building fast and effective back-testing systems which allow for executing and playing back in-sample and out-of-sample tests for strategies in seconds. The quality of our back-testing systems and proprietary algorithms allow matching 99% of the intra-day trades with back-tests. They support the following output parameters:

  • Cumulative profit over time
  • Average rate of return and Std of return over time
  • Predictive power
  • Winning trades/winning time percentage
  • Return ratio VS Risk
  • Sharpe ratio

DataArt is familiar with specifics of order execution in algorithmic trading such as:

  • Cancelling and amending orders
  • Hidden/Visible order
  • Aggressive/Passive execution
  • Order size strategy
  • Execution strategy

Our experience and knowledge of algo trading and data mining opens additional opportunities for those who use mathematical models and quantitative analysis for big data, web and social web mining.

Please provide your contact details to download the AIFMD application: submit an RFP form or send an email to sales@dataart.com.