The team at Borealis AI has built Prism, a simulated financial market that supports algorithmic trading on a fictional stock called “PRSM” – to help expose those interested in AI and finance to AI-based trading. The public competition gave participants a chance to battle it out and test their models in a safe, simulated trading environment on Prism.
How the competition worked
Borealis AI opened the Prism platform to interested participants who could use either their own homemade trading algorithms, starter models, or a manual trading interface, to compete and trade the fictional ‘starting wealth’ of $500. The leaderboard helped rank everyone on their realized profits.
It wouldn’t be a competition without some head-to-head action. And in the Prism competition, participants aimed to forecast and exploit one another’s actions, testing and perfecting their trading strategies and algorithms to come out on top. Throughout, the Prism platform evaluated the models and provided metrics on how players performed relative to one another.
Participants were also subject to certain position limits and budget constraints (they could not place bids that exceeded their wealth, for example). And while a manual trading interface was available for diagnosis and verification, participants were required to submit trades through an algorithmic trading agent. Testing environments were made available to allow participants to ensure their agent could interact with the Prism server.
Why we built our own stock exchange
We built Prism and hosted the Prism competition to showcase how AI-based trading environments work, help build financial literacy, and give more people in the wider ML community an opportunity to learn about AI and finance, in a hands-on, safe environment. At Borealis AI, we have a mission and deep drive to create real-world impact through scientific pursuit; Prism competition reflects our belief that exploration – while at times challenging, can also be fun.
“At Borealis AI, we’re passionate about our work – not just because it’s interesting and can be hugely impactful. But also, because we have fun and get to build exciting things,” “The Prism project ticked all the boxes.”
Meidan Alon
Sr. Machine Learning Engineering Lead, Borealis AI
“Building a stock exchange involves an exciting set of engineering challenges like low-latency order placement and data transfer, accuracy and fairness during trade execution, and high availability + fault-tolerance considerations. It's rewarding to see all the moving pieces come together to create a fun and educational experience for the ML community,”
Jawad Ateeq
Machine Learning Software Engineer, Borealis AI
Stay tuned on more Prism news, and thank you for participating in the first public competition (if you did). Please consider joining our talented team of engineers and researchers. Click here to explore the open roles at Borealis AI.
Explore