In an office overlooking downtown Boston, the views are partially obscured by math formulas and technical drawings that have been scribbled on the windows. Wearing a T-shirt with the words “machine earning” printed on it, Luca Lin says the formulas are being developed to help trade futures contracts and currencies at lightning-fast speeds.
“We are doing on average $1 billion of daily transactions,” says Lin, a co-founder of Domeyard. “It’s a high frequency trading strategy that is signal based.”
At 27, Lin is part of a new wave of entrepreneurial young quants who are starting trading firms in the face of daunting obstacles and challenges. But Lin and his two 26-year-old co-founders, Christina Qi and Jonathan Wang, have found believers.
Domeyard has raised $10 million for its general partnership from the likes of Howard Morgan, a co-founder of Renaissance Technologies who later became a venture capitalist, and Gary Bergstrom, the founder of quantitative investment firm Acadian Asset Management. Domeyard’s 14 employees include former portfolio managers who led high frequency trading teams at Quantlab, Athena Capital Research and Sun Trading, as well as former senior engineers from PDT Partners and Lime Brokerage—some of the biggest names in quantitative and high frequency trading.
Brash and optimistic, Domeyard’s founders have structured their firm as a hedge fund that doesn’t charge its investors a management fee, but does take between 40% to 50% of the profits. Qi says the firm, which currently manages in the low tens of millions of dollars, runs a low capacity strategy that currently makes between 10,000 to 40,000 trades daily. Although run as a hedge fund, Domeyard closes out its trades like many proprietary trading firms do, ending each day with no market exposure.
The Domeyard crew is operating in a field dominated by big firms with years of operating history that have spent fortunes on infrastructure and armies of mathematicians and engineers. In addition, this low-volatility stock market era has cut deeply into some of the richest strategies of high frequency traders, causing a wave of consolidation in the industry.
But Domeyard’s young founders think that there are some advantages to being the new kids on the high-frequency block. The firm is working to unlock profitable trading strategies by using sequential machine learning and making large scale computations of statistics. “I feel like we can do better in a lot of areas and with some technological problems because we started from scratch,” says Wang.
Wang was studying electrical engineering and computer science at the Massachusetts Institute of Technology when he started trading in his dorm with Qi, who lived in the room next to his. Wang had interned at Apple, but he found the technical challenges in trading intriguing. “The financial world is full of really tough technical problems,” Wang says. Qi had interned at Goldman Sachs and knew she wanted to pursue trading.
The duo hooked up with Lin, who was studying physics and math at Harvard University when he came across statistical arbitrage trading strategies while researching machine learning. “I had no idea what statistical arbitrage was but I realized they were using some of the same techniques I was using in biophysics,” says Lin.
Lin, who grew up in Singapore, started trading the European session in the early mornings with Qi and Wang at their MIT dorm, making about 2000 euros daily. After Qi and Wang graduated, they founded their firm in a Cambridge apartment. Lin joined them and they called their company Domeyard, a name that merged the landmarks of their schools—MIT’s Great Dome and Harvard Yard.
From the start, Domeyard’s founders divided their responsibilities clearly: Lin was responsible for the trading strategies, Wang for the technology and Qi for all the business functions, including fundraising. All three would pull night shifts babysitting the algorithms. The firm eventually moved to a WeWork space and then got its own office in downtown Boston.
It took nearly three years for Domeyard to get up and running. Qi was effective at raising money, but the young traders needed to build up the technology and infrastructure required for data-driven trading—securing servers in the Chicago Mercantile Exchange’s data center, writing a few million lines of code, and setting up the capability to support several petabytes of data. “We built all our high frequency tools internally, the servers, the colocation, the dark fiber channels and networks, all the software,” says Qi.
Domeyard has focused on trading U.S. stock futures and currencies with strategies that Lin claims are not predatory or seeking to be the fastest. He says the firm conducts huge numbers of statistical arbitrage trades seeking to make profits maybe 60% of the time. Lin further explains that Domeyard has worked at using sequential learning in a high frequency trading context to make statistical-driven trades that are not high conviction, essentially seeking out a small edge on each trade and then making a large number of trades. Domeyard searches for all sorts of sparse signals that could trigger a trade, like the speed or load of the market servers.
Domeyard turned their algorithms on in the first half of 2016 and the hedge fund started trading in September of last year. The firm appears to be making money. If the trading business is successful, Lin's overarching goal is to branch out to areas outside of financial services. Lin hopes that in the future Domeyard will find signals that can help target genes in biomedical research or maybe increase the language translation capabilities of computers. Says Lin: “We see high frequency trading as one application of the research and machine learning work we are doing.”
