Creative human prediction vs. the benefits of AI: in conversation with AlpacaJapan

In the complex world of FX trading, there is a constant demand for data, speed, and scalability, increasing accuracy and opportunity. AlpacaJapan Co. Ltd., a Tokyo-based fintech referred to as Alpaca, is working to meet this need by bringing scientific prediction into trading.

Expanding on their core mission, Alpaca built an AI Prediction Matrix, which delivers real-time horizon market forecasts utilizing an AI-based deep learning engine. These forecasts are available on the Bloomberg Terminal via the APP Portal (APPS ALPM <GO>).

Bloomberg recently spoke with Morifumi Yotsumoto, Alpaca’s co-founder and CEO, to discuss the company’s value proposition, how AI can contribute to FX trading workflows, and what is next for the financial landscape in APAC and other regions.

How have you seen artificial intelligence (AI) and machine learning (ML) shape the financial landscape in APAC over the past three to five years? How is it different from other regions?

At Bloomberg’s buy-side conference in New York last year, 9% of attendees indicated they were using AI as either a strategy or tool. When I attended the same conference as a speaker in Tokyo, about 8% of attendees raised their hands. This indicated to me that the level of adoption was about the same across different regions.

I believe these numbers will grow significantly in the next three to five years. When I look at our existing client base, banks and brokers are the most active sectors, while asset managers have been a bit slower to adopt this technology, due to their accountability to investors.

What are some of the data challenges that clients and the industry are facing when it comes to ML? How do they cope with them and what is the role of data quality and real time feeds in this process?

This is a very important point. In order to leverage ML for each kind of strategy, data engineering is one of the most important parts of the process. When we talk about data engineering, this includes data analysis, anomaly filtering, seasonality check, data cleaning, visualization, and so on. Without paying attention to governance and management, AI and ML systems cannot operate efficiently and apply their learnings long-term.

At Alpaca, we have a good data engineering team that operates separately from data scientists. In order to speed up the rate of innovation, we put a new focus on the data engineering team, with a unique mission that now encompasses everything but modelling. The purpose of this shift was to allow researchers to focus on the real value that Alpaca can provide through innovative models. This team now works closely together to ensure maximal productivity of the data scientists.

How has ML and data been applied to generate alpha for FX traders? Why was there a need to create something with that technology to help with this specific problem?

Candle charts provide a good example, since this kind of graphic visualization is ideal for human traders. They analyze these types of charts that essentially aggregate innumerous data points and the millisecond variations of each tick. AI and ML can easily interpret these individual data ticks as historical patterns and feed that data into systems at a granular level. FX traders might be able to identify the same patterns that an AI system can, but they have a difficult time predicting movements from big data and tick move. Human traders are able to have more time to be creative by using our apps, a collaborative platform between people and data science.

What were some of the factors that influenced the creation of a market forecast app that delivers real-time market forecasts?

Integrating with real time market data, such as bid-offer spreads, price distribution, mid-price move in tick level and tick counts, is key in making better trading decisions. If you are able to save money and time with every trade you execute, there are significant long-term benefits for corporations and their shareholders.

What are the benefits and challenges of using AI for forecasting FX movements?

It depends on each use case, but some are being used for new proprietary trading strategies, and others for enhancing the timing of TWAP (time-weighted average pricing) based hedges. There are also applications for digital automation of short-term trading. Our challenges and new opportunities are never-ending, to be honest. Models have to be continually updated, optimized and integrated until all of our customers are satisfied working with us. Ideally, this results in a win-win relationship.

How did being a part of Bloomberg’s app portal come about and what have the benefits of that collaboration been?

This relationship is 100% a joint effort. Since we launched the AI Prediction Matrix app together, we were able to tap into a large range of potential users globally, which would have never happened without the partnership. Bloomberg’s global solution and relationship with users optimized our platform globally, not only in Japan, but also in Asia, New York and London. There is still a lot of work to be done, but Bloomberg’s support in marketing and initiating our human-friendly AI tool has been invaluable.

What changes and innovations do you see in the near future when it comes to AI and ML shaping the financial landscape in APAC? Will it be different from other regions?

I think it is not only about Asia, but the global landscape in general. I hope many more AI and ML products for the financial markets are developed and that users can have more of a choice when they are deciding on who to work with. We have spent three years deep diving in tech discussions with professional institutional investors, which I’d say is our biggest advantage and edge. I believe that without human intelligence, good AI systems cannot be developed.

Going forward, our challenge is around AI automation. Leveraging more automation would allow us to focus more on solving complex problems and using cutting edge technologies. Our aim is to stay on the frontline of innovation and delivering high quality AI solutions to global customers with various financial products.