LINEAR REGRESSION

An automated trading firm turns its machine learning expertise to venture capital investing

XTX in London.
XTX in London.
Image: Stephen Bennett/Peldon Rose
By

XTX Markets doesn’t have any human traders. But it does have human venture capitalists.

XTX is well known in the financial industry as an automated trading firm—it buys and sells some $250 billion of assets, from foreign-exchange to stocks and US Treasuries, each day. It’s the only company in the top ranks of currency trading that isn’t a global investment bank, according to annual polls of institutional investors by Euromoney.

Much less is known about London-based XTX’s venture capital investments—XTX Ventures—which it hasn’t discussed publicly until now. Alexander Gerko, a Deutsche Bank alum with a PhD in mathematics from Moscow State University, founded London-based XTX in 2015. Zar Amrolia, a fellow math whiz who also worked at Deutsche Bank, joined the same year as co-CEO.

XTX executives say their trading edge is in machine learning—algorithms that automatically learn and improve from data without being re-programmed. XTX is heavy on computing, but light on personnel; the investment banking divisions of the world’s mega banks typically employ thousands, while XTX has a staff of about 140.

XTX Ventures focuses its investments on the machine-learning algorithms and tech that power the rest of the company. Its portfolio is made up of strategic investments in fintech and capital-markets startups, and venture investments in early-stage companies in any industry and in any location, from healthcare to advertising to robotics, according to Ekaterina Holt, who is head of the division. Gerko and Amrolia are on the investment committee, as are the company’s tech, quant, and regulation experts. She says the venture unit followed soon after XTX Markets was founded, and that the company is discussing its portfolio more publicly now that they have a “solid core of investments” across industries.

“We’re keen to show what we have invested in and what we are looking to invest in in terms of our investment thesis,” Holt said.

XTX didn’t disclose the assets under management of its venture unit or its returns over the years.

It’s not unheard of for financial companies and trading firms to make venture-capital investments. Jump Trading, a powerhouse based in Chicago, has a portfolio that spans tech, software, and media, according to its website. Susquehanna, a quantitative trading company, has invested in the likes of Credit Karma, a fintech, and ByteDance, operator of video-sharing app TikTok. Jim Simons, a pioneer of quantitative investing, also invested in tech startups at the same time that he was building his hedge fund Renaissance Technologies in the 1980s.

“It’s always a good idea to have some sort of VC unit to keep an eye on new technologies that might impact their business, and maybe there are ways to partner,” said Larry Tabb, head of market structure research at Bloomberg Intelligence.

“AI and ML are absolutely going to change the face of the world,” he added. “The challenge is betting on the right horse.”

Ekaterina Holt.
Ekaterina Holt.

Holt says investing in machine-learning startups is a chance to earn a financial return, of course, but it’s also an opportunity to see companies use the technology for a wide range of purposes.

She says the trading company’s machine-learning expertise and industry contacts are an advantage when it comes to buying stakes in startups. XTX uses its own capital for investing and doesn’t raise outside funds. “We leverage XTX Market’s industry expertise to do deep due diligence as well as help our portfolio grow,” she said. “We have an edge in understanding machine learning and internal resources to support companies beyond just cash.”

XTX appears somewhat unique for its especially narrow focus on machine-learning startups. It’s an area that’s expected to grow quickly—the commercial AI and ML market is worth about $63 billion, according to PitchBook data, and is forecast to grow 26% annually in the coming years.

 

Some think the Covid-19 pandemic has accelerated shifts toward automation, particularly in financial services, as companies look to cut costs and lean into virtual services. PitchBook senior analyst Brendan Burke wrote in a report earlier this year that artificial intelligence and machine learning, thanks to growing computing power and enhanced datasets, are only in the early stages of changing society through the replication of human intelligence.

Holt says the tech is helping solve a wide array of problems, from identifying cancer to containing the Covid-19 pandemic. “Of course financial return is a metric for us, however we are also interested in our portfolio companies deploying ML technology to solve specific problems across many areas like agriculture, healthcare, finance,” she said.