Nikita Ivanov, Founder & CTO, GridGain
Spread betting is an increasingly popular form of wagering. It allows traders to bet on whether the outcome of an event will be above or below a spread set by a bookmaker. Events can include changes in housing prices, the movement of a stock index, the difference between the two final scores at a sporting event, or anything else for which there is a measurable outcome that can go in either of two directions. Spread betting can be conducted through brokerages, asset managers, online gambling firms, and any other venue willing to set the spread and host the wager.
To be successful at spread betting, traders and venues must have high performance technology solutions. Those solutions must be able to stream data into models that quickly compute event relationships and change outcome probabilities while the events are occurring. To meet these technology challenges, spread betters are turning to in-memory computing. For a deep dive into this topic, read “Powering Financial Spread Betting with In-Memory Computing,” a new white paper by GridGain Systems, the leading provider of open source in-memory computing solutions for the financial services industry.
Spread betting currently offers a number of compelling advantages for both traders and venues compared to other forms of speculation:
- Low entry and transaction fees with a large upside potential
- Preferential tax treatment and less regulation
- A diverse array of products, markets and strategies
- Fully electronic platforms with single account access and mobile trading
- Automatic stop losses to help traders limit their losses
Spread betting also comes with a number of significant risks:
- Less consumer-protection oversight
- No “market maker” with large assets to assume the risk and step into a volatile market
- Wider bid/offer spreads and highly leveraged transactions, increasing the risks for traders
- High interest rates for overnight financing
- High volatility with significant price fluctuations, but low transparency into the bookmaker’s prices.
Spread betting traders and venues can limit their risks using a combination of the following strategies:
- Hedging. Hedges are alternative bets or investments that provide some protection in case of adverse movements in the original position. Developing a hedging framework typically involves mathematical or statistical models.
- Modeling outcomes. Predicting outcomes successfully typically results from building mathematical or statistical models based on outcomes of related events. Such models are complex and rely on feeding accurate data into probability regression analyses.
- Subscribing to data services. Subscription data services can provide the accurate and timely data needed for mathematical and statistical models.
- Analysing news. Current news can impact people’s opinions and change their betting strategy. Sentiment analysis involves analysing historical patterns in people’s opinions about certain types of news, then incorporating these patterns into the spread betting predictive models.
Implementing these strategies requires high performance technology solutions that enable traders and venues to stream data into models that quickly compute event relationships and changing outcome probabilities while the events are occurring. These technologies include:
- Big data. Interpreting large amounts of data from subscription services in real time requires big data technologies that organize large datasets into multiple pools and connect them for immediate analysis.
- Apache™ Hadoop® with MapReduce. Hadoop with MapReduce organizes hierarchical data to dramatically improve performance.
- Complex event processing (CEP). This technology can look at multiple streams of incoming data and use artificial intelligence (AI) to identify meaningful events. This is particularly useful for mathematical models that receive multiple streams from multiple event lines.
- Robots. When many people receive similar information and bet on the same assets, traders can gain advantages by building their own algorithms or “robots” to automatically buy or sell depending on the data the system receives.
- Online and cloud platforms and mobile trading. The spread betting industry typically uses mobile and Internet-based platforms that must be highly scalable and performant.
- Data partitioning and parallel processing clusters. Spread betting systems need additional performance improvements to handle the many different assets that are traded, as well as a wide variety of products. Data partitioning and parallel processing clusters are essential for supporting 24/7 data access and transactions. Parallel processing clusters also make it easy to scale the cluster as the number of users increases.
The key to performance: In-memory computing
While all these technologies contribute to high performance, a spread betting solution that depends on disk reads and writes – the slowest part of any computer system – will suffer from unacceptable delays. To overcome this, spread betting traders and venues are turning to in-memory computing, the fastest available storage-based computing method. Companies implementing in-memory computing in their spread betting systems have reported processing transactions roughly 1,000 times faster than disk-based solutions.
In-memory computing is a performance and scalability revolution driven by two key forces. First, the cost of memory has dropped roughly 30 percent per year since the 1960s, so it is now relatively inexpensive to equip clusters of machines with hundreds of gigabytes or even terabytes of RAM to support very large and very fast data projects. This in turn has led to a rapid maturation of the in-memory computing market. It used to be that to deploy even a bare-bones in-memory computing solution, organizations had to configure and cobble together multiple products. Today, however, full-featured in-memory computing platforms offering ACID-compliant transactions, transparent scalability, high availability, and enterprise-grade security for online analytical processing (OLAP) and online transactional processing (OLTP) are readily available to spread betting traders and venues. Organizations with an existing code base can even find an in-memory computing solution that includes SQL support and a full range of APIs, leading to much faster deployments.
As the popularity of spread betting continues to surge, the performance burden on systems will only increase. In-memory computing solutions can now deliver the performance that traders and venues require in an affordable platform.