Oxford University Multidisciplinary Researcher Develops World's First GPU Acceleration LOB Simulation

Oxford University illustration. (photo: twitter @UniofOxford)

The Oxford University recently developed a Limit Order Book (LOB) simulator accelerated by a GPU called JAX-LOB. This is the first in the world.

JAX is a tool to train the high-performance machine learning system developed by Google. In the context of the LOB simulator, JAX allows artificial intelligence (AI) models to train themselves directly on financial data.

The Oxford research team created a new method where JAX can be used to run a LOB simulator using only GPUs. Usually, LOB simulations are carried out using a computer processing unit (CPU).

By running it directly in the GPU series, modern AI training venues occur, AI models can go through several communication steps. According to the Oxford team's pre-print research paper, this provides a speed increase of up to 7 times.

The dynamics of LOB is one of the most scientifically researched aspects in the world of finance. In the stock market, for example, LOB allows full-time traders to maintain liquidity during daily sessions. In the world of cryptocurrencies, LOB is well received at almost all levels by professional investors.

Training AI systems to understand LOB dynamics is a difficult task and requires a lot of data, due to the nature and complexity of financial markets, depending on simulations. And the more accurate and powerful the simulation, the more efficient and beneficial the models it trains.

According to the Oxford team's paper, finding ways to optimize this process is very important. "Because of their central role in financial systems, the ability to accurately and efficiently describe LOB dynamics is very valuable. For example, this may allow financial companies to offer better services or could allow governments to predict the impact of financial regulation on financial system stability," said research team Oxfrod.

As the first in the world, JAX-LOB is still in its early stages. The researchers emphasized the need for further research in their paper, but some experts have predicted that this could have a positive impact in the field of artificial intelligence and financial technology.

"Soft equipment like JAX-LOB is interesting because it looks like a type of device that might be used by powerful artificial intelligence in the future to conduct its own financial experiments," wrote Jack Clark, co-founder of Anthropic.