The rise of the DIY quant conjures up images of pony-tailed nerds coding out of their mother’s basement for the vague prospect of winning some prize money. In fact, most are professionals in a range of industries – they just don’t want to work for a hedge fund.
“To some extent, traditional firms are constrained by their own employment models,” says Jonathan Larkin, chief investment officer at crowd-sourced hedge fund Quantopian. “If these firms wanted to copy our model, they would need to pay quants on a formulaic performance basis. Large firms cannot do this because it would create a two-tier system with employees who are not paid formulaically on performance and another segment that is."
Of the 15 amateur quants to win the right to invest between $1m and $3m of the $250m Point72 Asset Management CEO Steve Cohen committed to the platform’s algorithmic traders last year, not one works in financial services.
They come from the U.S, Spain, Columbia, China, India, Canada and Australia, and range from quants studying computer science at Cornell, to senior data scientists working for internet companies to mechanical engineers.
A growing band of hedge funds are launching competitions to try and unearth quants and sell finance careers to a limited supply of talent. Cohen has complained about the dearth of talent on offer to hedge funds, but the best and brightest of a community of 120,000 data scientists, quants and mathematicians would often rather work under their own terms.
“From our vantage point, we see a lot of people with the types of skills that would be welcomed in a lot of industries, particularly finance, but that don’t necessarily want to be tied down to one sector,” says Larkin. “To some extent, it’s reflective of the evolution of work – people want to move around.”
Last year, Cohen invested $2m in Quantopian and also promised $250m of his own money for the best algorithmic trading strategies. 15 have been chosen, and will begin investing $1-3m pots of money this quarter. This follows the firm’s decision to start managing external capital in its own hedge fund last year.
The prospect of managing Cohen’s money has brought out the competitive spirit in amateur quants around the world. This time last year, Quantopian had 80,000 members – now it has 120,000 people all locked in a Darwinian battle for dominance.
When Quantopian started out, its main carrot was offering top members the chance to become a ‘quant in residence’ – an elite status to lord over fellow members. Then it began offering $100k to top quants to manage for six months, allowing them to keep any profits.
The motivation for those managing Cohen’s money is not the prospect of employment, but cold hard cash. They get to keep 10% of the profits generated by their trading strategy.
Quantopian has evolved from a small tech firm started by John ‘Fawce’ Fawcett in 2011, which offered free seminars to for professional and amateur quants to a decidedly more professional set up.
It now has 53 employees, up from 40 in April 2016, and has spent the past year building out its own trading and data science teams. Larkin was previously global head of equities at Millennium Partners and held the same role at Bluecrest Capital Management before joining Quantopian in June last year.
It also hired Dragan Skoko from Fidelity Investments as head of trading in July. He now has two traders working under him and Larkin says that the plan is to hire more. Alisa Deychman was also named head of algorithm development in December. Quantopian needs this expertise to ensure it can successfully implement the trading strategies.
It has a selection criteria for its DIY quants to rival any tough job interview process. While competition heated up on the platform after Steve Cohen’s commitment, the vast majority of strategies never see the light of day.
Larkin says that the biggest mistake is “over-fitting”, namely where a strategy looks good on paper but doesn’t cut it in real markets. Testing this is one of the biggest parts of Quantopian’s selection process, says Larkin.
If a strategy makes it through the initial screening process, Quantopian makes sure the code can run successfully on its own for six months. Then it goes into an automated screening process and, finally, is monitored by the firm’s own quant team. Anyone who makes it through this is then subjected to a background test and finally an interview with the team.
Larkin says the benefits to users are that they keep all the intellectual property, Quantopian does all the “unglamorous” side of the business like, say, corporate actions and its trading team ensures “high quality execution”.
Regardless, the whole idea of uncovering quant talent that doesn’t typically gravitate towards finance is gaining momentum. Not only does Quantopian now have competition from the likes other crowdsourced quant hedge funds like Numerai and Quantiacs, but traditional firms are running their own competitions.
Citadel is offering $25k prizes to students to compete in 18 ‘datathons’ aimed at uncovering the best student quants and Two Sigma partnered with data science platform Kaggle, offering $100k for the best machine learning algorithms (and raising their awareness of jobs in finance). Maybe some will want to work for a hedge fund after all.