AI Developments in the Brokerage & Trading Space

By

Raymond Yung

calendar

May 8, 2023

clock

4

Min Read

An inside look into TC Labs - our R&D team

Since ChatGPT launched at the end of 2022, the internet has been ablaze with discussion around the powerful impact of artificial intelligence (AI). This chatter was amplified by the swiftly following launches of Google's "Bard" in February and Microsoft's Bing chatbot in March, but for many outside the realm of AI and development, it can be challenging to decipher the realistic truth from the hype and envision the potential of this technology. 

We decided to take advantage of our in-house AI experts to get to the bottom of what developments lie beneath this online buzz and what changes we anticipate it'll drive for the brokerage and trading space. We sat down with Mohamed Benjannet & Elvys Linhares Pontes from our Labs team who work collaboratively to build TC's latest analytics and algorithms that help today's self-directed investors form confident and educated decisions. 

What are AI chatbots like ChatGPT and what technology makes them possible? 

An AI chatbot is a conversational agent that uses artificial intelligence (AI) such as natural language processing (NLP)  and machine learning (ML) to provide natural, fluent, dialogue-based responses to user queries.

NLP allows the chatbot to understand the meaning of human questions, while ML trains the chatbot on large volumes of data and enables it to learn from previously collected conversations.

AI chatbots use large language models to generate their answers. In the case of ChatGPT, it is powered by the GPT (Generative Pre-trained Transformer) architecture, which is a type of deep learning model specially designed for tasks such as language translation and text generation. This language model is called GPT-3 in the case of ChatGPT and was trained on huge amounts of data. The new generation of  chatbots combines deep learning approaches, Natural Language Processing (NLP), advanced dialog management, and other techniques to train their models.

Unlike other chatbots, ChatGPT uses a reinforcement learning model based on human input, which allows it to produce more relevant answers. Despite its capabilities, results are not completely reliable and answers may contain factual inaccuracies. Care should be taken when using the results of a language model, especially in high-stakes contexts that match the needs of a specific use case.

There's a lot of hype online around how this technology will revolutionize how we work. To what extent do you believe this is true and where do you think we'll see the most impact? Are there any industries or roles that will be impacted first? 

AI has made tremendous progress in changing the way we live and work today, and will continue to do so exponentially. This transformation is typically seen as complementary rather than a substitution with general productivity most impacted. Some occupations like interpreters, translators and legal assistants have seen more significant transformation, with the barriers to entry drastically lowered. These trends lead us to believe white collar jobs, especially those with little on-the-job training or low tacit knowledge requirements will experience the greatest change. 

Moving forward, we expect AI disruption to bring the most value in roles requiring lots of information/data, computation power and objectivity such as office and administrative support, medicine, science, education, engineering, legal and finance.  Additionally, jobs based mainly on processing skills such as programming and writing skills will be heavily impacted while those emphasizing critical thinking will be less so.  This is because ChatGPT-like technology has the potential to take over processing tasks held by humans, such as question-answering, drafting reports, and grammar corrections to cite few of them. 

For better or worse, in education, we’ve seen the application of ChatGPT for accelerating school work.  While this can certainly be problematic, AI could offer significant support for teachers and students alike in gathering relevant information and sources in an effective manner for further exploration. Additionally,  in the medical field, we have already seen the application of robots in surgery and assisting medical professionals in providing diagnostics. In engineering, AI has sped up the R&D process with its ability to rapidly create and test a variety of prototypes.

In short, AI applications will change the way we have access to information by facilitating the recognition of relevant information to a query by providing complete and complex answers in a short and objective way. I believe this is just the beginning of a new technology era championed by A.I.

What is the mission of the "Labs" team? What type of products and analytics have been produced so far? What specific expertise or backgrounds are necessary for success in the "Labs" team?

As Trading Central’s think tank and R&D unit, our mission is to transform complex, unstructured big data into actionable insights that broaden existing capabilities to better support our customers. With the application of NLP, ML and quantitative research, these analytics are subsequently developed and transformed successfully into TC's award-winning lineup of embeddable tools. These innovative tools are in turn deployed to investors globally through the industry's leading online brokerages and financial institutions.

Labs’ flagship products include Market Buzz, Crowd Insight and Fundamental Insight. The first two are based on analyst-trained NLP algorithms nicknamed “Felix” and process close to 60K pieces of online content daily to identify the key topics and sentiment of a particular financial instrument.  Felix’s motto is "Read less, Know more", tackling infobesity for today’s investors and supporting timely, educated trade decisions.  As for Fundamental Insight, our objective is to provide a concise, high-level (yet scalable) view on how a stock is performing across a metrics of 20 different fundamental factors.  Similar to our NLP products, our goal is to bring clarity to our clients who are bombarded with too much complex information such as piles of company financial reports.

Our team members typically have strong AI and/or quantitative backgrounds, combined with programming skills. In general, we’re a bunch of motivated, intellectually-curious and friendly individuals that love our work.  We want to do our best while appreciating the results as well as the process.

In the AI "world", how is Trading Central’s Felix different from ChatGPT or Google Bard?

ChatGPT and Google Bard are chatbots that can read a request from a user and answer it. The request can be on any subject or any domain which is why we refer to them as “open-domain”. Both can be very helpful for tasks like drafting an Email, writing small code snippets or to generate a summary about a well known subject as long as users ask their question correctly. 

Since both models are limited to their training data, neither can answer questions about recent events. For investors, this is a serious limitation as it’s important to be informed of market changes to make the right decisions. This is why we developed Felix at Trading Central. 

Felix is an AI specializing in financial news and social media, and processed in real-time. Unlike ChatGPT and Bart, Felix cannot answer questions in open-domain, but it is a finance expert. Felix is reading finance and economic news in real time in order to be informed about the most recent information. Information gathered by Felix will be shown in Market Buzz and Crowd insight where investors could use this information to easily identify investment opportunities. In this manner, Felix allows users to identify the assets receiving news attention and then to identify investment opportunities in real time. Also Felix provides investors with the most discussed topics about an asset as well as insights into the market sentiment momentum. 

In summary, ChatGPT and Bart are trained to answer questions in open-domain, but they aren’t updated with recent news while Felix is finance- specialized AI which aims to process real time information in order to identify opportunities for investors and help them save time reading the news.

How do you think AI will impact the investing space?

AI is already changing the face of finance and is likely to continue as the technology is refined. Today AI is already used in fraud detection, risk management, portfolio management, automated trading and also for news and market sentiment analysis. Data availability and computation cost are the two main obstacles for wide use but they are slowly being overcome.

Investors who master the use of the AI will gain a major advantage in avoiding bias and making the right decisions, but it’s important to be properly educated and informed about AIs applications, limitations, and risks.

What are Labs' plans in this pursuit/endeavor?

We believe in the collective wisdom of the AI expert community. With an aim to contribute and collaborate with peers, we’ve partnered with two leading universities in France to produce several academic papers that will be shared with the AI and investment communities. 

Continuous innovation is the DNA of the Labs team and we have no shortage of exciting ideas in our research pipeline. Our ultimate goal is to make the full value AI accessible to our customers through Trading Central’s products in a user-friendly and intuitive manner. Aside from NLP, we’ve recently entered a new area of ML and are charging into this less-explored "territory" to further expand our product offering.  In parallel, we’re exploring how ML might offer enhancements to existing flagship products such as Technical Insight, Fundamentals and Nowcasting. 

To avoid cliches like “spoiler alert”, we’ll simply hint to keep an eye out for  upcoming Labs products!

You may also like...

Raymond Yung

Head of Trading Central Labs