Financial institutions such as banks, hedge funds, and mutual funds use quantitative analysis to make stock trades. An Investopedia article indicates, “Quantitative trading consists of trading strategies based on quantitative analysis, which rely on mathematical computations and number crunching to identify trading opportunities. Price and volume are two of the more common data inputs used in quantitative analysis as the main inputs to mathematical models.”

It is critical for financial services organizations to stay ahead of the competition and maintain maximum profitability when stock trading. To meet this goal, financial firms develop their own algorithmic trading models which are considered protected intellectual property that is not shared. The trading models use computers to analyze a mix of proprietary data, statistical and risk analysis, and external data.

Trading strategies were traditionally developed by financial quantitative analysts (quants) using ‘what if rules’ to determine the best and most profitable trading opportunities. Once the trading strategies were refined, the trading criteria was hard coded into computer programs used in making real-time stock market trades. Trading programs were often run from financial services data center computers using central processing units for the computation. The massive amounts of data to be processed placed a strain on data center infrastructure. In addition, quantitative analysts could not keep up with the analysis required to update their trading models to reflect the constantly changing market and economic conditions. Algorithmic trading was created to help financial service organizations meet today’s fast paced stock trading needs.

Evolution of algorithmic trading

Financial services firms are increasingly building highly automated algorithmic trading systems using artificial intelligence (AI) for quantitative trading analysis. According to SG Analytics, “Algorithmic trading accounts for nearly 60 – 73% of all US equity trading – data analytics in the stock market.”

Algorithmic trading involves building unique computer models which find patterns or trends that are not typically perceived by humans scanning charts or ticker (price) movements. The algorithms use quantitative analysis to execute trades when conditions are met. A simple example would be, if the price of oil hits $130 and the US Dollar declines 5% over the previous two weeks, then sell Oil and buy Gold in a 20:1 Ratio. Mathematical statistics such as standard deviation and correlation would be added to the model to determine when to execute a trade.

Machine learning (ML) is especially valuable in algorithmic trading because ML models can identify patterns in data and automatically update training algorithms based on changes in data patterns without human intervention or relying on hard-coded rules. According to a Finextra article, “With the hiring of data scientists, advances in cloud computing, and access to open source frameworks for training machine learning models, AI is transforming the trading desk. Already the largest banks have rolled out self-learning algorithms for equities trading.”

How cloud-based, GPU-accelerated AI meets algorithmic trading needs

The complexity and infrastructure requirements of algorithmic trading make it important for financial organizations to have partnerships with technology providers. Many of today’s algorithmic trading systems are powered by advances in GPUs and cloud computing.

Microsoft and NVIDIA have a long history of working together to support financial institutions by providing cloud, hardware, platforms, and software to support algorithmic trading. Microsoft Azure cloud, NVIDIA GPUs and NVIDIA AI provide scalable, accelerated resources as well as routines, and libraries for automating quantitative analysis and stock trading.

The partnership between Microsoft and NVIDIA makes NVIDIA’s powerful GPU acceleration available to financial institutions. Azure supports NVIDIA’s T4 Tensor Core Graphics Processing Units (GPUs), which are optimized for the cost-effective deployment of machine learning inferencing or quantitative analytical workloads. The Azure Machine Learning service integrates the NVIDIA open-source RAPIDS software library that allows machine learning users to accelerate their pipelines with NVIDIA GPUs.

Tools needed to create and maintain trading algorithms

In addition to Microsoft Azure Cloud solutions, Microsoft also provides tools that help developers and quantitative analysts develop and modify trading algorithms.

Microsoft Qlib

Microsoft Research developed Microsoft Qlib which is an AI-oriented quantitative investment platform containing the full ML pipeline of data processing, model training, and back-testing—it covers the entire auto workflow of quantitative investment. Other features include risk modeling     , portfolio optimization, alpha seeking, and order execution.

Microsoft Azure Stream Analytics

Microsoft Azure Stream Analytics is a fully managed, real-time analytics service designed to analyze and process high volumes of fast streaming data from multiple sources simultaneously. Azure Stream Analytics on Azure provides large-scale analytics in the cloud. The service is a fully managed (PaaS) offering on Azure.

Patterns and relationships can be identified in information extracted from various input sources and applications. Financial institutions can create, customize, or train algorithmic ML trading models using the combination of SQL language and JavaScript user-defined functions (UDFs) and user-defined aggregates (UDAs) in the Azure Stream Analytics tool.

Summary

Financial institutions using legacy data centers can no longer keep up with the massive amounts of data and analysis required for today’s fast-paced stock trading. Algorithmic trading using AI and ML that don’t require human analysis are becoming the norm for stock trading. Microsoft and NVIDIA provide advanced hardware, cloud, AI, and software solutions for algorithmic trading to meet the needs of the digital age.