Bond Trading systems are still far behind the curve in embracing Artificial Intelligence and Algo Trading tools in order to auto-execute trades or analyse the market situation and gain an edge in the business. Most of the trading is high touch and quite often still voice based. On the contrary other markets and in particular Equities carry out 80-90% of their trading volumes without any human intervention. The main reason for this is the simplicity of products in the Equities space compared to Fixed Income. For e.g. Ford Motor will have a single equity stock listed in the New York Exchange and so automating the trading would be quite straight forward. Whereas in Fixed Income markets, Ford Motor will have dozens and dozens of securities with different currencies, geographical entities, maturities, coupons and so on.
And that’s why even though the global Fixed Income market is much bigger than global equity markets in volume, most of the trading is manual. Even in this manual setup traders have very few tools to actually empower him to enable him to take accurate, fast and informed decision.
Having said that, Fixed Income is now moving fast towards automation. Post recent regulatory changes (MiFID II) most trading now has to happen via ECNs like Bloomberg, Tradeweb and MarketAxess. Bigger banks are quickly investing into Data Science Tools to untangle the complexity of Fixed Income products and monitor anomalies in the market in realtime and use it in their favour.
The entire world cutting across industries is moving towards data driven systems. Data collected historically enable us to take informed decisions. These data can be fed into AI systems that can produce results.
In this blog I will talk about the areas for upscaling Fixed Income Trading Systems to incorporate newer technologies and the solutions provided by CurveCue.
The fuel for AI systems is data. Do we lack data in Fixed Income trading?
The answer is NO. We have a lot of data. With the advent of big data and big processing power we can utilize these disparate data sets and build complex AI based solutions for Fixed Income trading. We have historical data of trades done by clients, we have inquiries data, we have news from various sources, we have several other internal data all of which can be normalised and fed into a complex AI system.
Do we need to port data out of bank systems?
The answer is NO. We can have the entire framework deployed within the bank. The framework of managing algorithms can take care of training, retraining and monitoring these AI based algorithms.
Is it very complex to build a system of managed algorithms?
The answer is NO. We can have a product build that can manage all these in an automated way without requiring intervention from the IT team of the bank.
We at CurveCue have built a Managed Algorithms module that abstracts away all the complexities and presents the algorithm through a simple REST interface. The module takes care of all the complexities of automatic training of models, monitoring and predictions.