TradingFinance - Petaflops

·         Blockchain
·         Wealth Management
·         High Frequency and Algorithmic Trading
·         Compliance
·         Equities and other Recommender Systems
·         Derivatives Pricing
·         Simulation
·         Analytics and Big Data – reducing the footprint and operating costs


High Performance Computing is essential for many aspects of Blockchain processing especially as volumes increase.
Blockchain’s distributed and decentralised storage of data offers scope for shared storage of other private information and introduction of smart contracts will have significant impact on transactions between organisations such as settlement.

Wealth Management

AI recommender systems can provide personalised recommendations to high net worth individuals. These systems take market intelligence, a complete analysis of individual portfolios and the historic investment strategy to customise recommendations. Not only are recommendation personalised but the speed of GPU-accelerated means that the frequency of these personalised recommendations can be significantly increased yielding higher levels of customer retention.
Many of the latest AI Voice to text to Speech systems are so effective that customers are not aware that they have been conversing with a machine. Low-latency AI conversions mean that systems can convert speech to text, interpret the text, research a response and translate this to speech in fractions of a second. As such real time personalised investment and portfolio recommendations are already possible.

High Frequency and Algorithmic Trading

High Frequency Trading has demanded High Performance Compute solutions for many years where the difference in performance of the network card and processor is critical to the business.
As we move into the era of Artificial Intelligence and Algorithmic trading systems high performance computing is more important than ever to gain an edge over your competitors.
In an effort to gain an edge many application developers are leveraging GPU-accelerated compute systems and other compute accelerator cards to reduce the processing time of critical functions within the trading applications. Many software developers have experienced a 10X-100X improvement in processing times.
We have high-performance GPU-accelerated systems featuring high-speed network cards commonly used in trading systems (like the Solarflare).

Equities and Recommender Systems

Machine learning based recommender systems for investment products can propose investment ideas in a similar way that Netflix makes recommendations on your viewing.
Recommenders can consider each investor separately and provide customised recommendations based upon their existing portfolio, risk profile, trading history, market volatility, current prices and more. What is more these recommendations can be made in real time considering the current news, pricing and current portfolio performance.
Many of these systems are based upon deep learning and continually improve the quality of recommendations.
These immediate, personalised and current recommendations improve service, frequency of client contact and build client loyalty. They can be combined with artificial intelligence for voice and speech processing to extend these systems to proactive contact with clients when alert conditions are triggered.

Derivatives Pricing

Deep Learning models have been developed to take pricing information and based upon training determine what happens next. These predictions are free of human emotion and bias and can be fed into options and other pricing models.
JP Morgan has adopted Deep Learning technology for Hedging finding the AI developed “deep hedging” strategies outperformed human-developed strategies. JP Morgan then began using the self-taught algorithms to hedge some of its vanilla index options portfolios. The bank plans to roll out Deep Learning technology for hedging single stocks, baskets and light exotics.

Simulation