Why hedge funds don't have a clue about how to use their quants, AI experts and data scientists
Quants and data analysts may be locked in a battle for dominance with discretionary hedge fund managers, but those in traditional roles may secretly know that their days are numbered.
“I’ve had billion dollar portfolio managers tell me that they were studying a data science course at night school,” said Leigh Drogen, a former hedge fund CIO who now runs big data fintech firm Estimize. “These courses weren’t even related to finance, because they don’t exist yet.”
“Most discretionary fund managers think Python is a big snake,” adds Emmett Kilduff, a former Morgan Stanley equity capital markets banker who now runs big data firm Eagle Alpha. “They’ve never heard of web scraping, or other big data techniques. The skill-set just isn’t there.”
To quants and data scientists, discretionary fund management is “religion” predicated on gut feeling and faith. But in the vast majority of hedge funds and asset managers, they still hold all the cards. The problem is that this ‘them and us’ mentality means no one wins.
“Most hedge funds have a centralised team, where you have an overarching group looking for data, ingesting it and cleansing it. Then you hire data scientists and quantitative researchers to look for alpha opportunities and basically hand over an Excel spreadsheet to the portfolio manager who has no idea what to do with it,” said Drogen speaking at the Newsweek Artificial Intelligence in Capital Markets conference today.
Drogen’s theory is that PMs, data scientists and quants should all work together in ‘pods’. “The PM understands the trading strategies, can explain his understanding of stocks and where he sees alpha to a quant and data scientist who can mine and ingest the data to give them an edge,” he said.
Most fund managers have yet to grasp this. More to the point, when you’re talking about a large firm, trying to bring in so much scarce data science talent within so many different teams is just not feasible, said Kiduff.
“A centralised quant team supporting traditional expertise is the way to go right now,” he said.
The whole idea of blending fundamental knowledge with huge quant datasets – the quantamental approach – is also gaining traction. But even here it’s not simply a case of gaining access to these huge sources of data, it’s a more question of what you do with it and that requires an understanding of the underlying financial drivers.
“The truth of the matter about the quant versus fundamental debate is no one has figured out not only how to use data science to gain an insight, but how to turn that insight into action,” said Michael Beal, CEO of big data focused hedge fund Data Capital Management. “Closing that loop and turning it into money is the hardest part.”
Despite the ongoing debate, a lot of portfolio managers in discretionary hedge funds know that the landscape has shifted and are taking action to update their skill-set, said Drogen.
“You have portfolio managers with decades of experience attending courses in Python and R, or learning how to build a factor model,” he said.
The shift towards big data and artificial intelligence in hedge funds is being held up those working in the big jobs right now. “Religion worked for a while,” said A.J. DeRosa, head of global sales at big data firm Orbital Insight. “You’re dealing with egos and you’re dealing with people, so you need to be empathetic. But in five years’ time, they either become quants or quantamentals or they won’t be around anymore.”
The whole idea of discretionary hedge funds hiring a bunch of data scientists and PhDs then sticking them in a back room to work their magic is something that needs to change. The cultural shift is likely to be gradual, but by the time it happens a new breed of fund manager may be way ahead of them anyway.
“There are around 70 hedge funds who say they use big data, about 20 of them are really doing in and maybe a handful are any good at it,” said Beal.
One of these is Numerai, the Silicon Valley-based hedge fund run by 29-year-old South African Richard Craib. It uses thousands of freelance data scientists to create machine learning models that are then used to make trades. There are around 13,000 people competing against one another to create the best strategies – the prize being around $150k in bitcoin.
“One of our investors Howard Morgan – co-founder of Renaissance Technologies – stopped investing in quant funds for a few years. Hi logic was how can you compete with Two Sigma, which has 120 PhDs using data sets in ways that no one else can?” he said.
“But we have seven or eight employees and 13,000 data scientists all around the world building the hedge fund without us. We’re in a third wave and a new kind of hedge fund is being created.”
Photo: Getty Images