Ultimate magazine theme for WordPress.

Toss of a coin that made a one-time game developer top of the quants

0 15

When Igor Tulchinsky was deciding whether to join hedge fund Millennium Management three decades ago, the Belarusian former video game programmer eschewed the systematic, data-driven approach that is characteristic of “quants” like him. He simply flipped a coin.

It was not “so that chance shall decide the affair, while you’re passively standing there moping”, to quote a poem by Danish scientist Piet Hein. “But the moment the penny is up in the air, you suddenly know what you’re hoping.”

The coin toss dictated that Tulchinsky should stay at his current employer, options trader Timber Hill. But when he announced the decision to Millennium, “I felt so bad,” he recalled in a rare interview. So he changed his mind and quit Timber Hill for Millennium.

His about-face turned out well for Millennium’s investors. Tulchinsky became one of its top portfolio managers, and in 2007 spun out a quantitative investment manager, WorldQuant, to manage money for the now roughly $60bn-in-assets hedge fund group. Over the past 15 years WorldQuant has grown into one of the largest and highest-contributing units at one of the world’s top hedge funds. People familiar with the matter say it now manages more than $10bn, split between the $7bn or so it trades on behalf of Millennium and a roughly $3bn fund open to other investors.

New York-based Millennium is a so-called multi-manager fund, which allocates money to about 290 different teams of traders across a wide range of investment strategies.

The fund is up around 10 per cent this year to the end of November and has gained an average of 14 per cent a year since it was founded in 1989 by Izzy Englander, according to investors. WorldQuant’s own 15-year record is shrouded in secrecy — even to its own employees and Millennium investors — but insiders say Tulchinsky is one of the highest-paid people in the wider group.

WorldQuant operates in a field of quantitative investing known as “statistical arbitrage”.

Its model is to produce algorithms that try to predict the price movements of various financial instruments, typically equities, and then take advantage of inefficiencies in the markets. Tulchinsky, whose intense gaze and head-to-toe black attire give him the air of a James Bond villain, refers to these algorithms as “alphas” and to WorldQuant as the “Alpha Factory.”

“This whole business grows at an exponential rate, and the data itself is growing exponentially,” he said, outlining how WorldQuant’s own growth has coincided with the exponential growth of data. “We believe the amount of data produced in the next five years will be the same as the amount of data ever produced . . . Today we have 1,500 data sets that we combine in all kinds of ways.”

In a 2017 paper, Tulchinsky memorably depicted technology as a “Quantasaurus” that stalked slow-moving prey. “The Quantasaurus has the potential to create far more than it destroys,” he wrote. “The Quantasaurus will survive, but if we choose to pen it up or flee from it, we will surrender many of its benefits.”

The problem confronting many quants is that what was once an “alpha-rich” environment — dominated by unsophisticated stock brokers and old-fashioned mutual funds that were easy pickings for sophisticated players — has become much trickier to navigate. Quant firms find themselves in an unending battle to sniff out new signals from trading patterns, find new data sets, explore new approaches and mine any inefficiencies before their rivals do.

“Trading signals decay, whether you use them or not, because if you don’t use them others do,” said Tulchinsky. WorldQuant has a library of more than 5mn signals but he estimates that each signal decays — that is, loses its predictive power — by an average of 15 per cent each year. “In the other direction we keep coming up with new signals so it’s a never-ending cycle.” He adds that while “crowding is a concern . . . as long as we just keep researching fast enough, keep finding new signals fast enough, diversifying in every possible way, then crowding is more of an intellectual than a practical concern”.

For WorldQuant, this diversification includes an expansion into new markets, including high-frequency trading and options trading, and trying to create an “Alpha Factory” in corporate bonds — a new frontier for investors. Its headcount has roughly tripled over the past decade, to more than 800 people (compared to more than 4,500 at Millennium), and the firm is one year into a three-year plan to significantly increase staff numbers.

At the same time, it is moving beyond inputting economic and corporate data into its models and using artificial intelligence to look at whether it is possible to predict in real time what future data points will look like — an approach Tulchinsky dubs “data 3.0”. 

“Data may be easier to predict than stock movements because your activity might influence the stock movements but your activity does not influence the cash flow of some company,” he said.

It is also exploring whether it can speed up its trading and compete more directly with high-frequency trading firms to make money out of more fleeting market inefficiencies. The idea is “for now” not to set up a broker-dealer like rival Two Sigma and try to compete on pure speed, but to see if WorldQuant can “create higher-frequency alphas just like we can medium-frequency alphas”, Tulchinsky said.

His has been an unlikely path, even for a corner of the investment industry with more than its fair share of unconventional origin stories. Born in Minsk, Belarus, to parents who were professional musicians, Tulchinsky began playing chess as a child and discovered computer programming at middle school. He started developing video games when he was 17 years old.

“I think being a refugee from the Soviet Union and having seen my parents take a risk and come to the US puts risk-taking EQ [emotional quotient] in my head,” he said. “I’m comfortable with risks, I like making fast decisions.” Coin flips often help decide the trickier ones.

In the early 1990s, when Tulchinsky was a young trading strategist at Timber Hill looking for a job, he sent thousands of letters to company chief executives. “For every 1,000 letters you get about 10 interviews and two job offers, but only if you send to CEOs and only if you flatter them in a vague way. If you send 1,000 letters to Wharton alumni [where he did an MBA] you’ll get a thousand letters back with pages of thoughtful advice but no job offer.”

Tulchinsky’s unorthodox path to the top of the hedge fund industry has informed his approach to building WordQuant, where the geographical and decentralised distribution of its workforce is unusual. It is built on the premise that “talent is distributed equally around the world, opportunity is not”, said Tulchinsky. “And we provide opportunity to the talent.” 

WorldQuant’s headquarters is in the hedge fund heartland of Old Greenwich, Connecticut, but its offices in 13 countries are spread across many non-traditional financial centres, such as Ramat Gan, Israel; Budapest; Mumbai; Ho Chi Minh City and Seoul. “The idea is that if you want to hire the smartest people in the world, they can’t all be in New York City.”

He launched WorldQuant University in 2015, a philanthropic venture that is a free online university offering masters degrees in financial engineering and applied data science courses to students all over the world. And it has built a separate division called WorldQuant Predictive — “AI on demand” — which sells predictive analytics to corporate clients.

After bringing in retired US Army general Stanley McChrystal as a consultant, every two weeks WorldQuant’s entire workforce joins a “giant Zoom call, carefully choreographed”. Tulchinsky is also fond of using anonymous surveys, sometimes several a day, because they “give you the ability to take any idea and pose it out there to 800 smart people and maybe have it shut down in approximately one minute”, he joked.

Right now, the war for skilled staff is “quite extreme”, he said. Most of the people who leave WorldQuant go to high-tech start-ups rather than rival hedge funds.

Tulchinsky said that staff turnover at WorldQuant is 5-10 per cent a year. But at least one high-profile hire has proved shortlived: Gary Chropuvka, the former co-head of quantitative investment strategies at Goldman Sachs, joined WorldQuant as president in 2020 but left after just over a year.

Everything comes back to Tulchinsky’s business philosophies, including “quantity is quality” and “everything is information”. WorldQuant uses crowdsourcing to try to improve performance at the margin. This year it launched the WorldQuant Brain platform that allows people to build and submit “alphas” for potential compensation and it holds regular competitions that it sees as a pipeline for talent.

“Crowdsourcing may be great, but you wouldn’t use crowdsourcing to design a spaceship,” said Tulchinsky. “We already have a successful business . . . we already have a system that determines whether something’s valuable or not, so we’re just looking for incremental improvements.” 

Leave A Reply

Your email address will not be published.