Author Archives: Sarah

  • -

Point 72 Invests $250M in Algo Platform

B_08Point 72 Invests $250M in Algo Platform

Point72 Asset Management’s Steven A. Cohen, the billionaire investor operating under a two-year ban from managing others’ money, is putting $250 million of his own funds under management with Boston-based startup Quantopian, and has bought a $2 million piece of ownership in the algorithmic trading platform.

Read the full article

  • -

The Rise of the Artificially Intelligent Hedge Fund | Wired

by CADE METZ | JAN 25, 2016

LAST WEEK, Ben Goertzel and his company, Aidyia, turned on a hedge fund that makes all stock trades using artificial intelligence—no human intervention required. “If we all die,” says Goertzel, a longtime AI guru and the company’s chief scientist, “it would keep trading.”

He means this literally. Goertzel and other humans built the system, of course, and they’ll continue to modify it as needed. But their creation identifies and executes trades entirely on its own, drawing on multiple forms of AI, including one inspired by genetic evolution and another based on probabilistic logic. Each day, after analyzing everything from market prices and volumes to macroeconomic data and corporate accounting documents, these AI engines make their own market predictions and then “vote” on the best course of action.

Though Aidyia is based in Hong Kong, this automated system trades in US equities, and on its first day, according to Goertzel, it generated a 2 percent return on an undisclosed pool of money. That’s not exactly impressive, or statistically relevant. But it represents a notable shift in the world of finance. Backed by $143 million in funding, San Francisco startup Sentient Technologies has been quietly trading with a similar system since last year. Data-centric hedge funds like Two Sigma and Renaissance Technologies have said they rely on AI. And according to reports, two others—Bridgewater Associates and Point72 Asset Management, run by big Wall Street names Ray Dalio and Steven A. Cohen—are moving in the same direction.

Automatic Improvement

Hedge funds have long relied on computers to help make trades. According to market research firm Preqin, some 1,360 hedge funds make a majority of their trades with help from computer models—roughly 9 percent of all funds—and they manage about $197 billion in total. But this typically involves data scientists—or “quants,” in Wall Street lingo—using machines to build large statistical models. These models are complex, but they’re also somewhat static. As the market changes, they may not work as well as they worked in the past. And according to Preqin’s research, the typical systematic fund doesn’t always perform as well as funds operated by human managers.

In recent years, however, funds have moved toward true machine learning, where artificially intelligent systems can analyze large amounts of data at speed and improve themselves through such analysis. The New York company Rebellion Research, founded by the grandson of baseball Hall of Famer Hank Greenberg, among others, relies upon a form of machine learning called Bayesian networks, using a handful of machines to predict market trends and pinpoint particular trades. Meanwhile, outfits such as Aidyia and Sentient are leaning on AI that runs across hundreds or even thousands of machines. This includes techniques such as evolutionary computation, which is inspired by genetics, and deep learning, a technology now used to recognize images, identify spoken words, and perform other tasks inside Internet companies like Google and Microsoft.

The hope is that such systems can automatically recognize changes in the market and adapt in ways that quant models can’t. “They’re trying to see things before they develop,” says Ben Carlson, the author of A Wealth of Common Sense: Why Simplicity Trumps Complexity in Any Investment Plan, who spent a decade with an endowment fund that invested in a wide range of money managers.

This kind of AI-driven fund management shouldn’t be confused with high-frequency trading. It isn’t looking to front-run trades or otherwise make money from speed of action. It’s looking for the best trades in the longer term—hours, days, weeks, even months into the future. And more to the point, machines—not humans—are choosing the strategy.

Evolving Intelligence

Though the company has not openly marketed its fund, Sentient CEO Antoine Blondeau says it has been making official trades since last year using money from private investors (after a longer period of test trades). According to a report from Bloomberg, the company has worked with the hedge fund business inside JP Morgan Chase in developing AI trading technology, but Blondeau declines to discuss its partnerships. He does say, however, that its fund operates entirely through artificial intelligence.

The whole idea is to do something no other human—and no other machine—is doing.
The system allows the company to adjust certain risk settings, says chief science officer Babak Hodjat, who was part of the team that built Siri before the digital assistant was acquired by Apple. But otherwise, it operates without human help. “It automatically authors a strategy, and it gives us commands,” Hodjat says. “It says: ‘Buy this much now, with this instrument, using this particular order type.’ It also tells us when to exit, reduce exposure, and that kind of stuff.”

According to Hodjat, the system grabs unused computer power from “millions” of computer processors inside data centers, Internet cafes, and computer gaming centers operated by various companies in Asia and elsewhere. Its software engine, meanwhile, is based on evolutionary computation—the same genetics-inspired technique that plays into Aidyia’s system.

In the simplest terms, this means it creates a large and random collection of digital stock traders and tests their performance on historical stock data. After picking the best performers, it then uses their “genes” to create a new set of superior traders. And the process repeats. Eventually, the system homes in on a digital trader that can successfully operate on its own. “Over thousands of generations, trillions and trillions of ‘beings’ compete and thrive or die,” Blondeau says, “and eventually, you get a population of smart traders you can actually deploy.”

Deep Investing

Though evolutionary computation drives the system today, Hodjat also sees promise in deep learning algorithms—algorithms that have already proven enormously adept at identify images, recognizing spoken words, and even understanding the natural way we humans speak. Just as deep learning can pinpoint particular features that show up in a photo of a cat, he explains, it could identify particular features of a stock that can make you some money.

Goertzel—who also oversees the OpenCog Foundation, an effort to build an open source framework for general artificial intelligence—disagrees. This is partly because deep learning algorithms have become a commodity. “If everyone is using something, it’s predictions will be priced into the market,” he says. “You have to be doing something weird.” He also points out that, although deep learning is suited to analyzing data defined by a very particular set of patterns, such as photos and words, these kinds of patterns don’t necessarily show up in the financial markets. And if they do, they aren’t that useful—again, because anyone can find them.

For Hodjat, however, the task is to improve on today’s deep learning. And this may involve combining the technology with evolutionary computation. As he explains it, you could use evolutionary computation to build better deep learning algorithms. This is called neuroevolution. “You can evolve the weights that operate on the deep learner,” Hodjat says. “But you can also evolve the architecture of the deep learner itself.” Microsoft and other outfits are already building deep learning systems through a kind of natural selection, though they may not be using evolutionary computation per se.

Pricing in AI

Whatever methods are used, some question whether AI can really succeed on Wall Street. Even if one fund achieves success with AI, the risk is that others will duplicate the system and thus undermine its success. If a large portion of the market behaves in the same way, it changes the market. “I’m a bit skeptical that AI can truly figure this out,” Carlson says. “If someone finds a trick that works, not only will other funds latch on to it but other investors will pour money into. It’s really hard to envision a situation where it doesn’t just get arbitraged away.”

Goertzel sees this risk. That’s why Aidyia is using not just evolutionary computation but a wide range of technologies. And if others imitate the company’s methods, it will embrace other types of machine learning. The whole idea is to do something no other human—and no other machine—is doing. “Finance is a domain where you benefit not just from being smart,” Goertzel says, “but from being smart in a different way from others.”

Source: Wired

  • -

Artificial intelligence: Can Watson save IBM? |

by Richard Waters | January 5, 2016

Hopes pinned on developing system into a powerful and profitable set of services

The history of artificial intelligence has been marked by seemingly revolutionary moments — breakthroughs that promised to bring what had until then been regarded as human-like capabilities to machines.

The AI highlights reel includes the “expert systems” of the 1980s and Deep Blue, IBM’s world champion-defeating chess computer of the 1990s, as well as more recent feats like the Google system that taught itself what cats look like by watching YouTube videos.

But turning these clever party tricks into practical systems has never been easy. Most were developed to showcase a new computing technique by tackling only a very narrow set of problems, says Oren Etzioni, head of the AI lab set up by Microsoft co-founder Paul Allen. Putting them to work on a broader set of issues presents a much deeper set of challenges.

Few technologies have attracted the sort of claims that IBM has made for Watson, the computer system on which it has pinned its hopes for carrying AI into the general business world. Named after Thomas Watson Sr, the chief executive who built the modern IBM, the system first saw the light of day five years ago, when it beat two human champions on an American question-and-answer TV game show, Jeopardy!

But turning Watson into a practical tool in business has not been straightforward. After setting out to use it to solve hard problems beyond the scope of other computers, IBM in 2014 adapted its approach.

Rather than just selling Watson as a single system, its capabilities were broken down into different components: each of these can now be rented to solve a particular business problem, a set of 40 different products such as language-recognition services that amount to a less ambitious but more pragmatic application of an expanding set of technologies.

Though it does not disclose the performance of Watson separately, IBM says the idea has caught fire. John Kelly, an IBM senior vice-president and head of research, says the system has become “the biggest, most important thing I’ve seen in my career” and is IBM’s fastest growing new business in terms of revenues.

But critics say that what IBM now sells under the Watson name has little to do with the original Jeopardy!-playing computer, and that the brand is being used to create a halo effect for a set of technologies that are not as revolutionary as claimed

“Their approach is bound to backfire,” says Mr Etzioni. “A more responsible approach is to be upfront about what a system can and can’t do, rather than surround it with a cloud of hype.”

Nothing that IBM has done in the past five years shows it has succeeded in using the core technology behind the original Watson demonstration to crack real-world problems, he says.

Watson’s case

The debate over Watson’s capabilities is more than just an academic exercise. With much of IBM’s traditional IT business shrinking as customers move to newer cloud technologies, Watson has come to play an outsized role in the company’s efforts to prove that it is still relevant in the modern business world. That has made it key to the survival of Ginni Rometty, the chief executive who, four years after taking over, is struggling to turn round the company.

Watson’s renown is still closely tied to its success on Jeopardy! “It’s something everybody thought was ridiculously impossible,” says Kris Hammond, a computer science professor at Northwestern University. “What it’s doing is counter to what we think of as machines. It’s doing something that’s remarkably human.”

By divining the meaning of cryptically worded questions and finding answers in its general knowledge database, Watson showed an ability to understand natural language, one of the hardest problems for a computer to crack. The demonstration seemed to point to a time when computers would “understand” complex information and converse with people about it, replicating and eventually surpassing most forms of human expertise.

The biggest challenge for IBM has been to apply this ability to complex bodies of information beyond the narrow confines of the game show and come up with meaningful answers. For some customers, this has turned out to be much harder than expected.

The University of Texas’s MD Anderson Cancer Center began trying to train the system three years ago to discern patients’ symptoms so that doctors could make better diagnoses and plan treatments.

“It’s not where I thought it would go. We’re nowhere near the end,” says Lynda Chin, head of innovation at the University of Texas’ medical system. “This is very, very difficult.” Turning a word game-playing computer into an expert on oncology overnight is as unlikely as it sounds, she says.

Part of the problem lies in digesting real-world information: reading and understanding reams of doctors’ notes that are hard for a computer to ingest and organise. But there is also a deeper epistemological problem. “On Jeopardy! there’s a right answer to the question,” says Ms Chin but, in the medical world, there are often just well-informed opinions.

Mr Kelly denies IBM underestimated how hard challenges like this would be and says a number of medical organisations are on the brink of bringing similar diagnostic systems online.

Applying the technology

IBM’s initial plan was to apply Watson to extremely hard problems, announcing in early press releases “moonshot” projects to “end cancer” and accelerate the development of Africa. Some of the promises evaporated almost as soon as the ink on the press releases had dried. For instance, a far-reaching partnership with Citibank to explore using Watson across a wide range of the bank’s activities, quickly came to nothing.

Since adapting in 2014, IBM now sells some services under the Watson brand. Available through APIs, or programming “hooks” that make them available as individual computing components, they include sentiment analysis — trawling information like a collection of tweets to assess mood — and personality tracking, which measures a person’s online output using 52 different characteristics to come up with a verdict.

At the back of their minds, most customers still have some ambitious “moonshot” project they hope that the full power of Watson will one day be able to solve, says Mr Kelly; but they are motivated in the short term by making improvements to their business, which he says can still be significant.

This more pragmatic formula, which puts off solving the really big problems to another day, is starting to pay dividends for IBM. Companies like Australian energy group Woodside are using Watson’s language capabilities as a form of advanced search engine to trawl their internal “knowledge bases”. After feeding more than 20,000 documents from 30 years of projects into the system, the company’s engineers can now use it to draw on past expertise, like calculating the maximum pressure that can be used in a particular pipeline.

To critics in the AI world, the new, componentised Watson has little to do with the original breakthrough and waters down the technology. “It feels like they’re putting a lot of things under the Watson brand name — but it isn’t Watson,” says Mr Hammond.

Mr Etzioni goes further, claiming that IBM has done nothing to show that its original Jeopardy!-playing breakthrough can yield results in the real world. “We have no evidence that IBM is able to take that narrow success and replicate it in broader settings,” he says. Of the box of tricks that is now sold under the Watson name, he adds: “I’m not aware of a single, super-exciting app.”

To IBM, though, such complaints are beside the point. “Everything we brand Watson analytics is very high-end AI,” says Mr Kelly, involving “machine learning and high-speed unstructured data”. Five years after Jeopardy! the system has evolved far beyond its original set of tricks, adding capabilities such as image recognition to expand greatly the range of real-world information it can consume and process.

Adopting the system

This argument may not matter much if the Watson brand lives up to its promise. It could be self-fulfilling if a number of early customers adopt the technology and put in the work to train the system to work in their industries, something that would progressively extend its capabilities.

“Once it’s working, you want to be leading the adoption,” says Shaun Gregory, head of technology and strategy at Woodside. “You’re ahead in knowledge and learning. Machines never forget,” he says.

Another challenge for early users of Watson has been knowing how much trust to put in the answers the system produces. Its probabilistic approach makes it very human-like, says Ms Chin at MD Anderson. Having been trained by experts, it tends to make the kind of judgments that a human would, with the biases that implies.

In the business world, a brilliant machine that throws out an answer to a problem but cannot explain itself will be of little use, says Mr Hammond. “If you walk into a CEO’s office and say we need to shut down three factories and sack people, the first thing the CEO will say is: ‘Why?’” He adds: “Just producing a result isn’t enough.”

IBM’s attempts to make the system more transparent, for instance by using a visualisation tool called WatsonPaths to give a sense of how it reached a conclusion, have not gone far enough, he adds.

Mr Kelly says a full audit trail of Watson’s decision-making is embedded in the system, even if it takes a sophisticated user to understand it. “We can go back and figure out what data points Watson connected” to reach its answer, he says.

He also contrasts IBM with other technology companies like Google and Facebook, which are using AI to enhance their own services or make their advertising systems more effective. IBM is alone in trying to make the technology more transparent to the business world, he argues: “We’re probably the only ones to open up the black box.”

Even after the frustrations of wrestling with Watson, customers like MD Anderson still believe it is better to be in at the beginning of a new technology.

“I am still convinced that the capability can be developed to what we thought,” says Ms Chin. Using the technology to put the reasoning capabilities of the world’s oncology experts into the hands of other doctors could be far-reaching: “The way Amazon did for retail and shopping, it will change what care delivery looks like.”

Ms Chin adds that Watson will not be the only reasoning engine that is deployed in the transformation of healthcare information. Other technologies will be needed to complement it, she says.

Five years after Watson’s game show gimmick, IBM has finally succeeded in stirring up hopes of an AI revolution in business. Now, it just has to live up to the promises.

Key assets: Systems depend on diverse data flows

There is a secret weapon in the race between leading tech companies to create the most effective forms of artificial intelligence: access to large amounts of data. For IBM, as it tries to make Watson a new standard for AI in the business world, this could turn out to be an under-appreciated advantage, according to some experts.

At the heart of intelligent machines are algorithms that search through large volumes of data to identify patterns and make surmises. Machine learning — the basic technique behind many of the recent advances in AI — relies on using large amounts of data to train systems in this way.
“A lot of what is now emerging with AI technologies has to do with data,” says Kris Hammond, a computer science professor at Northwestern University.

One of the biggest reasons for recent advances in AI has been the availability of large amounts of data online with which to train systems. Google has perfected its search systems by harnessing the abundant data it collects about the online behaviour of its users. While IBM cannot match the massive trove of information Google has at its disposal, what it lacks in volume it hopes to make up for in industry-specific detail.

“Google has one kind of data — consumer sentiment data. We have a vast amount of [more diverse] data,” says John Kelly, head of research at IBM.

The more industry-specific data it is fed, the smarter it will become at solving business problems. As customers pour their own corporate information into Watson in order to train it, IBM stands to be a beneficiary.

Watson’s “expanding corpus of information in many domains” could turn out to be one of IBM’s main assets in the AI race, says Mr Hammond.

Last year, IBM turned to acquisitions to boost its reserves of data. These included the $1bn it spent to buy Merge Healthcare, a company that handles large amounts of medical images. It has been folded into Watson Health, the first industry-specific business unit to be spun out of the Watson division.

It also spent $2bn to buy the digital assets of the Weather Company, with the aim of feeding its weather data into forecasting systems geared to understanding weather-related business risks, among other functions.

“Between our customers and what we’ve acquired, we’re amassing quite a data set,” says Mr Kelly.

Spinning this basic raw material into computing gold still requires serious technical skills. But if it can persuade customers to contribute their own data to the task of making Watson smarter, it could deliver the sort of head start that will make it hard for rivals to catch up.

Source: Financial Times

  • -

The New Quant Hedge Fund Master

by Nathan Vardi | JAN 4, 2016

This story appears in the January 18, 2016 issue of Forbes.

On a recent rainy October evening, Peter Muller, 52, sits at a piano on the stage of Manhattan’s City Winery, playing with the band from his third album, Two Truths and a Lie. In between songs about love, heartbreak and relationships, like his Bruce Hornsby-reminiscent “Kindred Soul,” Muller describes the long, strange trip he has taken in and out of high finance.

The tieless suits in attendance, from places like Goldman Sachs and Blackstone, paid as much as $1,000 a ticket to raise nearly $55,000 for the Robin Hood Foundation. And while Muller tells them of his early discovery of music, the existential crisis of his 30s, buddies he left behind in California and his family, there is a sense that many in the room just want to be in the orbit of the hottest hedge fund manager on Wall Street today. “I know you all had your choice of hedge fund manager CD-release parties,” quips Muller. “Thank you for choosing ours.”

Pete Muller is the latest, greatest member of a growing band of hedge funds that use complex math and computer-automated algorithmic models to buy and sell stocks, futures and currencies based on statistical correlations and aberrations that can be found in the market. During 2015, when many hedge fund managers–from mighty activists like Bill Ackman to noted short-sellers like David Einhorn–lost money, Muller spun the market’s volatility into gold. The largest fund of his three-year-old PDT Partners firm, which oversees $4.5 billion, was up 21.5% net of fees in the first 11 months of 2015.

“We knew that Pete has a magic touch,” says J. Tomilson Hill, the billionaire who runs Blackstone’s $70 billion hedge fund investment unit. “I happen to be a big fan of Cézanne, and Pete is in his own way as gifted as Cézanne was.” Paul Tudor Jones, the billionaire hedge fund manager, adds: “He is up there with the best and brightest–bar none.”

Indeed, Muller’s fund is so coveted that even Wall Street’s power elite are willing to effectively grovel to get in on PDT’s action. Many hedge funds stipulate that limited partners remain “locked up,” or prevented from redeeming funds, for a predetermined period, usually one year. PDT is the opposite. Its biggest investor, Blackstone, actually agreed to be locked up for no less than seven years–in return for Muller’s assurance that he would not kick it out of his biggest fund for the same period of time. Others requested the same lockup restrictions–and were refused. Even more astonishing is Muller’s 3% of assets under management fee, and performance fees that rise to 50% of profits for benchmark-beating performance, compared with the already maligned industry standard of 2-and-20.

“Our goal is to be the best quantitative investment firm on the planet, but not in terms of number of assets, in terms of quality of the products,” says Muller in his first interview since opening his new firm. “To take money out of the market with as little risk as possible and build a place people who are smart are drawn to.” Muller’s niche formula has also let him take plenty of money out of the market personally: Forbes estimates that in the last three years alone he’s made $200 million before taxes, including gains on his own capital.

It’s mid-November 2015, the U.S. stock market has given back all of its gains, and hedge fund managers around the globe are wringing their hands in anticipation of sending out another batch of disappointing investor letters. Muller is sitting in his Manhattan office. He is wearing a gray sweatshirt and jeans and has a Zen-like calm. His research chief has just left his office after telling Muller about a promising finding that could lead to the improvement of one of PDT’s main models. “When people buy or sell in a desperate or hurried fashion, it tends to be helpful to us,” says Muller, who is otherwise tight-lipped about what has gone right this year.

There are two screens in Muller’s office: a flat-panel display on his desk showing the movement of his hedge fund’s positions and a much larger screen on the wall that displays a real-time high-definition stream of the surfing beach at the foot of his house just north of Santa Barbara, Calif. When he’s at PDT’s headquarters in New York City and the waves are big, Muller sometimes yearns to be hanging ten. Luckily this doesn’t happen too often, because Muller spends two-thirds of his time at his California home, where responding to “surf’s up” is a regular ritual.

Muller may not appear to be a workaholic like many other Wall Street titans, but he is obsessive about his algorithms and problem solving, and he can get lost in deep thoughts for hours, days. His fear of burnout is real–he already dropped out of Wall Street once in 1999–and diversions like music and surfing are almost a necessity.

Muller grew up in suburban Wayne, N.J. His father was an electrical engineer and his mother a psychiatrist. He was good at numbers and loved music. At Princeton he studied math and played in a jazz band. After graduating, he headed to northern California to play music for rhythmic gymnasts and, figuring he had to pay the bills, eventually went to work for BARRA, a pioneering research firm that catered to quantitative financial firms. In 1992 he joined Morgan Stanley in New York as a proprietary trader to see if he could use math and computers to trade himself. Some of his investment banking colleagues were skeptical about the new math guy in the office. He called his group Process Driven Trading, or PDT. “I wanted to win and prove myself,” Muller says.

Nobody outside the bank knew it, but for a long time Muller was Morgan Stanley’s supersecret weapon, making big contributions to its earnings each year, hidden in the firm’s income statement under “principal transactions.”

Muller was able to carve out his own quiet area at Morgan Stanley’s Manhattan headquarters, where his team of math nerds could dress casually away from the bank’s testosterone-fueled, high-octane trading hordes. Muller became intensely focused on figuring out patterns that could help him beat the market. It was thrilling and exhausting. He thought and talked about it all the time–couldn’t even sit through a Broadway show without stressing over it. “He is really smart, but a lot of smart people get lost in theory,” says Kim Elsesser, a computer programmer and mathematician from MIT, and Muller’s first key hire. “He also has very high expectations of himself and other people.”

As Muller gained success and autonomy at Morgan Stanley his behavior became somewhat erratic. He detached from the office at a second home in Westport, Conn. in part because the pressures of work were overwhelming. His mind became so overloaded with mathematical formulas that he could no longer play music. Crossword puzzles became an escapist obsession; he even created them for the New York Times. By 1999 Muller started to feel like he could no longer find happiness on Wall Street.

“I was out of balance personally,” Muller says. He went on sabbatical, rediscovering his love of music partly by busking in New York subway stations and sojourning in far-off places like Bhutan. After returning in 2000 he spent the next several years essentially as an advisor to the fund he created, PDT. Muller today likens it to a kind of executive chairman position that left him time to do other things, such as practice yoga and produce two music albums with titles like Just One Lifetime. He also met his wife, Jillian.

The soul-searching lasted about seven years, and Muller says it sent his trading operation into a period of stagnation. Muller then rolled up his sleeves and came back full-time to PDT in 2006. Unfortunately his return just about coincided with the quant meltdown of 2007, when the precipitate drop in subprime mortgage securities triggered deep losses for many firms. Under pressure from Morgan Stanley, Muller was forced to liquidate part of his portfolio. “Morgan Stanley Star Is Among Those Battered; No Time for Music Now,” the Wall Street Journal ‘s front page blared.

As with many on Wall Street, the financial crisis changed the game for Muller. He had produced the kind of returns that would have made him a billionaire had he been an independent hedge fund manager. But working for Morgan Stanley always appealed because he didn’t have to worry about raising cash, appeasing clients or back-office details. It was plug and play. He couldn’t invest his own money in PDT, but he was well-paid, receiving a cut of his unit’s profits, and could singularly focus on solving market puzzles.

There was also the tricky issue of the intellectual property Muller developed but Morgan Stanley owned. But 2008 exposed the danger of being dependent on one client, namely Morgan Stanley. It also gave birth to the Volcker Rule, a piece of legislation designed to make it impossible for a proprietary trader like Muller to work at a bank like Morgan Stanley. Over the next few years Muller engaged in on-again, off-again negotiations with the Wall Street firm about their operating arrangement.

“We preferred to stay together, but as the Volcker Rule emerged it became clear that would not be permitted,” says Jim Rosenthal, Morgan Stanley’s chief operating officer, who led the last round of negotiations with Muller. “Sadly, this was a business that was a steady source of revenue and profitability and did not pose significant risks to the firm.”

In the end Muller would manage Morgan Stanley money until the end of 2012 and control the intellectual property Morgan Stanley was no longer permitted to use. Under the terms of the deal Morgan Stanley would get a cut of the fee revenue of the new, independent PDT for an undisclosed period of time.

On New Year’s eve 2012 Muller transferred all of his group’s investment positions from Morgan Stanley to PDT Partners. It wasn’t only the positions and intellectual property that came with him–so did every single member of his 80-person staff. Invigorated, Muller went to work, increasing his new business and nearly doubling his employees.

“It feels great to have your own place,” says Muller from his office on the top floor of a midtown Manhattan building formerly occupied by Random House. “I never felt like I had to have my name on the door, but I didn’t own it before, and in hindsight I didn’t recognize the psychological impact of that.”

In order to make his mathletes more comfortable Muller has had special glass walls constructed that are slightly curved to deflect sound and maintain the quiet workplace needed for concentration. Outside those quiet areas there are Ping-Pong and foosball tables near the kitchen and meeting rooms with whiteboards covered with mathematical formulas. Employees never wear suits; they run book clubs and organize poker nights that Muller sometimes attends. (He has made a final table at the World Series of Poker.)

Not much is known about Muller’s black box models. He traded using two different strategies at Morgan Stanley that have morphed into the two hedge funds he now runs. The PDT Partners Fund is a statistical arbitrage fund built on models that have never had a down year. The $3 billion fund was up 21.5% in the first 11 months of 2015, and given its high fees, its gross returns were running at about 40%. Since inception in 2013 PDT Partners Fund has produced annualized net returns of 18.5%. Another fund, $1.5 billion Mosaic, has a longer time horizon and had produced returns of 10.5% net of fees through November of last year and 8.5% annualized in three years. PDT also has a Fusion Fund, which allocates cash between PDT Partners and Mosaic.

Returns like that are beginning to rival the long-reigning king of quants, Renaissance Technologies, known for market-defying consistency and for producing a net worth of $14 billion for its professorial founder, James Simons. While Muller is not yet even a billionaire, some say he is the new Simons.

Like Renaissance PDT is a Ph.D. farm, with 35 researchers who spend most of their days developing trading algorithms. They are organized into five teams by the asset class and time horizons they work on. Some work on futures contracts with longer holding horizons while others toil with statistical arbitrage strategies that trade stocks over a medium time horizon of several weeks. Most efforts to come up with new models tend to start with two-week-long deep dives but can grow into research projects that last a year. PDT has one open problem today that its people have worked on for four years.

And while most Wall Street research analysts expect their best ideas to find their way into firm portfolios within weeks or months, PDT takes an academic approach to portfolio change. Researchers know that their models may not affect returns for two years or more. In fact, PDT is still using models today with concepts that were initially developed 15 years ago, but models do decay over time and need to evolve with the market. “We are more intent in building a group of Ferraris than a bunch of Toyotas,” says Tushar Shah, research chief at PDT.

Finding the right minds for Muller’s model-making is almost as hard as decoding statistical arbitrages hidden in markets. Big data and sheer computing power have become a driving force in PDT’s business model. Like other quants, PDT routinely competes with tech firms for leading programmers and mathematicians. It is now hiring more computer engineers than mathematician-researchers. Experts in machine learning are in high demand, so poaching talent from the likes of Google and Microsoft has become popular of late.

It’s not always the eye-popping first-year salaries of several hundreds of thousands of dollars that hook new Ph.D. recruits. PDT researcher John Sun, 30, was finishing up an MIT Ph.D. in electrical engineering and computer science when he got an e-mail from Eunice Baek, Muller’s longtime partner who manages recruiting. Sun would frequently get these sorts of e-mails and ignore them, but this one pulled him in. It said people at PDT like Lord of the Rings, science fiction and board games such as Settlers of Catan.

Like most PDT job candidates, Sun was flown to New York for a 36-hour interview on the second Saturday of November at PDT headquarters, where Muller and his partners tried to determine if Sun had the smarts and was someone with whom they could spend a lot of time. The guts of the recruiting weekend include a modeling interview and then collaborative algorithmic-based problem-solving games in which candidates are separated into small teams that Muller watches closely. PDT has a 3.5% turnover rate, and while the hours are not grueling, the work is demanding–trying to solve stock market puzzles often ends in failure.

“I want their shower time because in the shower they are thinking about things that get them to solve the problems,” says Muller.

While Muller is supersecretive about the details of his models even to limited partners, who seem to invest on faith–he is adamant that PDT does not engage in ultra-high-frequency trading. Ferreting out small market inefficiencies is core to PDT’s strategy, and what is also clear is that, for Muller, the more trading going on in markets the better. “There is a limitation on how much volume we can trade naturally built into our systems,” Muller says, adding he will almost certainly return some capital to his investors at the end of 2015, as he did in 2014. Trust in Muller’s machines is paramount, and he rarely intervenes manually, even when jarring upheavals temporarily defy his model’s predictions.

Spending seven months of the year with a surfboard at the ready or composing in front of a keyboard, instead of obsessively staring at a CNBC ticker, probably gives Muller’s PDT an advantage. Instantaneous information and constant volatility are the new reality of global markets. Whether it is index investing or robo-advisors, the discipline and brainpower of machines are winning on Wall Street. The rise of the quants is just beginning. “It will get harder, but we are prepared, and as information becomes more widely available and computing power increases, the strength of our models will improve,” Muller says. “Quantitative investing is the best way to manage money, period.”

Source: Forbes

  • -

Man vs. Machine in Trading and Investing

by Ruzbeh Bacha, Founder and CEO of CityFALCON | Sep 29, 2015

During the London Fintech Week last week, I had the opportunity to discuss this topic on a panel along with UCL professor Donald Lawrence and John Booth of Midpoint. Also, I have had long discussions on this subject with Niall Bellabarba who also follows this topic closely and recently published his own views.

Let me start by saying that I’m in no way expert on the topic, but I’m someone who’s trying to keep a tab on what’s happening on this space, and I would like to take advantage of emerging technologies at CityFALCON. Also, while I’ve got inputs from several people, the below are my personal thoughts on the topic. Moreover, I’ve kept it simple for any person without technical knowledge or with limited finance knowledge to understand.

If you trade like a machine, time to pack your bags might come soon

Several traders just follow the price action, regression testing, signals, etc, in trading. These tasks are programmable, and hence, machines can and are taking these over. Yes, some traders will talk about intuition and experience, but not sure if that provides them with an advantage.

Traders with coding abilities have an edge

Even though machines can take over algorithmic trading, banks will need coders to program the machines. But instead of hiring 10 traders on the floor, you now need 1-2 traders to work with the coders. If you can trade and code, then you’re a star. Interesting article on the subject – Why Goldman Sachs is replacing traders with coders.

Machines suffer from incompetencies of humans for now

Tesla’s April Fools joke was traded upon, and some firms would have lost money. Some claim machines are responsible for this move and 1 April was not built into the algorithms. Who should you blame in this case? Man or machine?!

Investing could also be disrupted by machines

Typically, it is not easy to invest relying only on financial ratios. You need to assess the market opportunity, quality of management, etc, before making your decisions. For example, you are unlikely to buy growth companies such as Facebook, Linkedin, and Twitter relying only on the price/earning ratio.

However, for some investing strategies, e.g. buying companies quoting below net market value of assets on their balance sheet or buying high dividend companies, machines could be easily trained.

Would you hire difficult-to-work-with traders or work with error-free machines?

Hiring and managing traders is expensive and a not-so-easy operation. Also, humans are likely to make mistakes that could bring down firms, e.g. with ‘fat fingers’. Machines, on the other hand, do not need breaks and do not suffer from fatigue, stress, depression, etc.

What happens in the lesser known world of artificial intelligence

If you don’t know what AI is, watch this video. When we get into a world with AI, where machines have the capability of a human brain, even subjective investing can be taken over by machines. Machines could get to a stage when they can say: I love Elon Musk – he’s is a visionary, and so, I’ll buy Tesla. A recent study by MIT shows that there are three categories of jobs where humans can perform better than machines: 1) Creative endeavours, 2) Social interactions, and 3) Physical dexterity and mobility. Unfortunately, these categories may not apply to trading and investing.

In summary, in the short term, the impact of machines could be limited to technical trading and some investing strategies, but in the long run, the market could be run just by machines. Who will have “an edge” then? The ones with more processing power? Also, what happens to retail investors? Could they take the benefit of trading through machines? What sort of regulation will we need?


  • -

MIT Quant Guru Andrew Lo on Market’s Meltdown: ‘August Sucks’

by Michael P Regan | Tuesday, September 8, 2015

Lo sees ‘a number of different forces’ coming to a head
Event risk is something algo trading ‘can’t really manage yet’

Andrew Lo has spent a lot of time peering into Wall Street’s various black boxes and “modeling the endogenous risk among hedge fund strategies.” The finance professor at Massachusetts Institute of Technology’s Sloan School of Management and chairman of AlphaSimplex Group LLC shared his thoughts on Friday about the recent spate of volatility in the stock market and what role strategies such as risk parity, trend-following commodity trading advisers and volatility targeting may have played.

Question: What does this volatility look like to you? Is this another quant meltdown?

Lo: I’m not sure I’d characterize it as just a quant meltdown. I think that makes it a little bit too cut and dried. Probably there are a number of different factors, including algorithmic trading, that plays into it. We have a number of different forces that are all coming to a head. And because of the automation of markets and the electronification of trading, we’re seeing much choppier markets than we otherwise would have five or 10 years ago. But it’s many forces operating at different time scales, all coming to a head.

Question: Is systematic trading exaggerating the moves?

Lo: I think it’s doing two things. One it can be exaggerating the moves if it lines up with what the market wants to do. So if the market is looking to sell because of an impending recession, then I think we’re going to see a lot of the algorithmic trading going in the same direction. And if the time horizon matches, you will see that kind of cascade effect. At the same time, I think algorithmic trading can play the opposite role. They can dampen some of the market swings if they’re going opposite to the general trend… The one thing that is true, though, is that algorithmic trading is speeding up the reaction times of these participants, so that’s the choppiness of the market. Everybody can move to the left side of the boat and the right side of the boat now within minutes as opposed to hours or days.

Question: When you talk about exaggerating the effect, is that mostly CTAs and momentum players or is it not that simple?

Lo: I think that over the course of the last few weeks, that’s actually a pretty decent bet: That there are trend followers that are unwinding because of some underperformance and concerns about the change in direction of the market. But, for example, what happened in August 2007 was equity market neutral strategies that unwound. So I think it really varies depending on the nature of the strategies that are getting hit and the money going into and out of those strategies, and how that’s affecting market dynamics.

Question: A lot of focus has fallen on risk parity strategies. The notion that, as volatility picked up, there was a lot of deleveraging going on, especially with futures and ETFs. Does that make sense to you from what we’ve seen?

Lo: Well, it certainly looks that way. Part of the challenge of risk parity is that it ignores anything about expected returns. The idea behind risk parity is not a bad one, which is to focus on risk and to manage your portfolio so as to try to stabilize that risk. But the problem with equalizing it across all asset classes or investments is that not all investments are created equal at all points in time. So there are certain strategies that end up doing worse than others during periods of times. And if you end up equalizing your volatility across those strategies, you might end up getting hit pretty hard as some of the equity risk parity strategies got hit over the course of the last few weeks.

Question: Is risk parity looking like a crowded trade?

Lo: I think there’s definitely a case in point of the idea of alpha becoming beta. The idea that once you start popularizing a particular investment approach, and it becomes so popular, that in and of itself creates these kinds of shock waves. So for example if the strategy itself underperforms, now we have a larger number of investors that are going to be unwinding that strategy and that will create a kind of cascade effect where the strategy will underperform even more as people start to take money out of the strategy. There are a number of examples. Risk parity, of course, is the most recent. But before that trend following, before that value investing, growth investing, earnings surprise, earnings momentum, any kind of a strategy can become a crowded trade. And when it does you have to just make sure that the risk premium associated with that trade is commensurate with the potential risks of getting hit with these unwinds.

Question: Are volatility targeting strategies part of the story? Have they become so popular that they’re exaggerating the moves?

Lo: Not only are they exaggerating the moves, but I think they are creating volatility of volatility. So it’s making the market quite a bit more complicated and the dynamics now are much more different and much more difficult to manage if you’re not aware of how these dynamics play out.

Question: What about when you get a big rebound? What do you suppose that is? Is that actually value-type of investors seeing the drops and coming in, or is it just another systematic trading function?

Lo: These rebounds are a confluence of a number of phenomena. One, you’re seeing that once selling pressure declines, investors will naturally become more optimistic and will come back into the market. That’s a common phenomenon. But I think that a rather newer phenomenon is the fact that these algorithms, because they operate at such high frequencies, when the price moves beyond a certain threshold, the algorithms will kick around and flip and go the other way. It’s happening at a rate that’s faster than it’s been anytime in the past because we haven’t had the technology to be able to do that.

And finally what we’re seeing is expectations shifting more rapidly because unlike five or 10 years ago we now have very big players in the financial markets, actively trying to move markets. In particular, I’m thinking about central banks and governments that are trying to manage economies by engaging in quantitative easing or other kinds of financial market transactions. When you have a small number of very big players that are going to be trying to move markets for political or long-term economic reasons, it becomes much, much harder to understand what’s happening. So people are all sort of trigger happy when small pieces of information hit the market, they tend to start moving money very quickly and in large size.

Question: Is that type government intervention something that algos can’t anticipate? Is that sort of an Achilles heel of algo strategies?

Lo: Absolutely. That event risk is something most algorithmic trading strategies really can’t manage yet. I say “yet” because in five or 10 years maybe natural language processing and artificial intelligence will have allowed them to read the news, interpret it and make judgments the way George Soros or Warren Buffett can. But I think we’re still a few years away from that

Question: Are a lot of momentum strategies able to turn on a dime that quickly? We’ll see this intraday drop of several hundred points, then it turns on a dime…

Lo: I think that it’s hard for momentum strategies to be able to move that quickly. In fact, some of the strategies that do move that quickly end up getting whipsawed. The real challenge in operating in these markets is that risk management would have you cut risk in the face of losses. The problem is that if you cut risk too quickly and by too much, you may end up missing out on the rebound, in which case you’ve locked in your losses and you might be getting back in the market exactly at the worst time. So you’re getting hit on both ends. What this atmosphere creates is a much more complicated challenge to risk managers to figure out what is the right frequency with which they need to cut risk and put it back. And I think everybody is trying to figure out what that optimal frequency is. But until we get a sense of who’s involved in the markets and driving these frequencies, it’s going to be anybody’s guess. And as a result a lot of people are going to be surprised over the next few weeks and months.

Question: Any other observations you have from the last couple of weeks that you think people might be interested in?

Lo: Yeah. August sucks.


  • -

Flash Boys Welcome: World Exchanges Woo High-Frequency Firms

by Sam Mamudi, John Detrixhe, Benjamin Bain | Sunday, 12 July 2015

Hey there, flash boy.

Exchanges around the world are avidly wooing high-frequency traders, those controversial speed demons of Wall Street.

Despite the often explosive debate over this kind of trading in the U.S., bourses in Mexico, Turkey, South Africa and beyond are trying to lure HFT types to boost business.

The message is clear: whatever the perceived risks, algorithmic robot traders — algobots — are marching steadily across the globe.

“We are welcoming foreign investors, and that includes HFT firms,” says Muammer Cakir, managing director at Borsa Istanbul.

For many exchanges, the attraction is obvious. Algobots buy and sell faster than a blink, boosting overall trading volumes. Higher volumes, in turn, help attract investors, creating a virtuous circle that benefits the exchanges.

To its advocates, high-frequency trading is nothing short of a revolution, making trading faster, cheaper and more efficient. They’re the modern market makers, replacing the men in blazers standing by their posts. To its detractors — among them author and journalist Michael Lewis, who added “flash boys” to the Wall Street lexicon — HFT lets savvy practitioners profit unfairly at others’ expense. U.S. authorities have voiced concerns and some are investigating the industry.

Many exchanges are unbowed. So far, high-frequency traders have a negligible presence in Istanbul. But Cakir’s exchange plans to upgrade its stock trading systems in September in the hope of attracting more HFT business, and it plans the same for its derivatives platform next year.

Close Computing

Borsa Istanbul also is letting traders place their computer servers next to exchange systems -– a practice known as co-location -– to facilitate HFT trading. The ultimate plan is for more institutional and foreign trading, and it’s betting that HFT activity on its venues will help make that happen.

The Tokyo Stock Exchange is taking similar steps. TSE officials last month visited New York to let the HFT industry know about upgrades due in September to its Arrowhead trading engine. Arrowhead already matches orders more than 1,000 times faster than was possible five years ago.

Tokyo has been helping high-speed traders in other ways. In the past two months, Susquehanna International Group and KCG Holdings Inc., gained “remote membership” at Japanese bourses, the first approvals since authorities opened up the possibility in 2009. The distinction allows firms to place buy and sell orders directly on the exchange.

The Hunt

Critics of high-frequency trading in the U.S. and Western Europe say that the fragmentation of those markets into multiple venues has allowed the algobots to hunt for profitable trades across several platforms, a strategy coined latency arbitrage.

In places like Tokyo, by contrast, exchanges are still largely monopolies. More than 90 percent of Japanese stock trading takes place on the TSE, for instance. And the single market hasn’t kept HFT out, which makes up about 44 percent of transactions.

The dominance of a single exchange in many Asian markets may have helped high-frequency traders avoid some of the controversies that have dogged them in the U.S.

“There hasn’t been this historical antipathy toward HFTs as being predators,” says Michael Syn, head of derivatives at Singapore Exchange Ltd.

Ill Will

Still, there’s plenty of ill will to go around. The algobots are fast replacing human traders, and in the U.S., high-frequency trading today accounts for at least half of all stock trading. Critics say HFT not only gives its practitioners an unfair advantage but can also unsettle markets. If the algobots pull out of a market all at once, their retreat might exacerbate sudden swings.

Various U.S. agencies have examined high-frequency trading but have found no evidence of widespread wrongdoing. This week, U.S. officials released a report concluding that in the Treasury market, HFT firms contributed to wild price swings on Oct. 15, 2014.

The exchanges, for their part, say hyperkinetic high-frequency trading ultimately narrows the differences between the prices at which people buy and sell and also improves trade execution. That in turn encourages trading by long-term investors.
“They’re not necessarily bad,” says Eduardo Flores, vice president of market supervision at Comision Nacional Bancaria y de Valores, which regulates Mexico’s financial system. “They’re market makers and in a certain form they help it so there’s more liquidity.”

In Mexico, the bourse is trying to attract more high-frequency traders to boost volumes, said Luis Carballo, the top information technology official at the Bolsa Mexicana de Valores SAB, which operates the exchange.

Mexico’s Wants

About 70 percent of the stock trades on Mexico’s national exchange already involve HFT, Carballo estimates. He said the Bolsa is analyzing how to make its market data platform more efficient.

“What we want is more trades,” he says. “We prepare the system, the technology, so that the possibility of doing high-frequency trading exists.”

Mexico exchange executives flew to New York last month for one-on-ones with clients of local brokerages, crisscrossing Manhattan to meet with algorithm coders, traders and others who use the Bolsa’s market data. They also met with officials from other exchanges to share ideas and expertise, according to Alfredo Guillen, the chief operating officer for equities at the exchange.

“We went specifically to speak about market data,” Guillen said in a telephone interview from Mexico City.

Meetings recently have become more focused on connectivity for traders as high-frequency trading has grown more common, according to Guillen.

Trading Catalyst

JSE Ltd., the company that operates the Johannesburg Stock Exchange, opened a co-location facility in May 2014 that cut the time it takes for data to travel from a trader to its servers and back to 150 microseconds, from 2,550 microseconds. Stock transactions rose 19 percent last year at the exchange and in October it had a record month, with daily average volume close to 400,000, about a third higher than its previous best.

“As part of our revenue-growth strategy, we engage in business origination where we seek new flow from a variety of market participants — buy side, sell side, retail as well as firms that employ low-latency trading strategies, also known as HFT players,” said Donna Oosthuyse, director of capital markets at JSE.

For more, read this QuickTake: Trading on Speed

Perhaps the last big obstacle to high-frequency trading achieving global dominance is China, where tight government rules, a stamp duty on stock trades and market inefficiency have so far kept out the algobots. There may be signs of opening up, though: Doug Cifu, CEO of Virtu Financial Inc., one of the world’s biggest computer trading firms, said on an earnings call in May that Virtu was having “very significant preliminary discussions” about entering the Chinese markets.


  • -

Artificial intelligence is the next big thing for hedge funds seeking an edge

by Georgia McCafferty | Monday, 4 May 2015

The glitzy skyscrapers of Hong Kong’s financial center stand in stark contrast to a dirty grey industrial building in the city’s run-down Kwun Tong district. Yet nine floors up, in an office bereft of any form of signage, a new artificially intelligent investor is taking shape.

This trading robot, developed by a team of academic roboticists, mathematicians, and ex-bankers, is the brainchild of fledgling hedge fund Aidyia, which has been seed-funded by venture capitalist Emanuel Breiter. Scheduled to start trading US equities this year, the team hopes it will deliver returns on the three years and millions of dollars it has taken to build, by turning huge swathes of financial and linguistic data into unique investment strategies.

“It’s seeing patterns that aren’t easy for the human mind to wrap itself around,” said Aidyia’s co-founder and chief scientist, Ben Goertzel.

Computer-assisted trading is nothing new. But Aidyia and other firms—including hedge fund giants Bridgewater Associates and Renaissance Technologies—hope to create intelligent software that can teach itself to adapt to changing market conditions, without guidance or instruction from humans.

Financial traders may not be the first people Elon Musk, Bill Gates and Stephen Hawking were thinking of when they expressed their recent concerns about AI’s threat to the human race. But if companies like Aidyia succeed, the human trading profession could well be one of its first casualties.

From algorithms to machines that can learn

Last year more than 40% of new hedge funds were “systematic,” meaning they used computer models for the majority of their trades, according to data provider Preqin—the highest percentage ever.

The closely-related field of algorithmic trading is designed to react extremely quickly to market changes. The algorithms seek out and exploit small windows of trading opportunity, often measured in minute fractions of a second. So many orders on the US stock market are now placed by automated algorithms that the Securities and Exchange Commission is looking for ways to regulate them as it does the rest of Wall Street.

These algorithms may work at superhuman speeds to identify tiny windows of trading opportunity, but ultimately they do exactly what they are programmed to do by humans.

Not so the AI systems. One of the quintessential things that sets apart the AI systems being designed for financial trading is their ability to learn and adapt.

Most quantitative trading, as it is currently practiced, relies on a human being to develop a mathematical model to identify trading opportunities. The model is then updated by hand to adapt to new markets or changing conditions. For an AI, conversely, humans develops the initial software, but the AI itself develops the model and changes it over time.

The trading robot developed by Aidyia ingests vast amounts of information, including news and social media, and uses its reasoning powers to recognize connections and patterns in the data. It then uses those patterns to make predictions about the market, which it translates into buy and sell orders—all without any direct human involvement.

Silicon Valley’s artificial brain drain

The latest advances in applying artificial intelligence to financial trading have been fueled by Silicon Valley, where companies like Google have invested heavily in machine learning to enable projects like self-driving cars. Bridgewater’s AI team is being led by David Ferrucci, who formerly managed the IBM team that developed Watson, the computer “Jeopardy!” champion that recently also created recipes for a cookbook.

Goertzel spent years researching and applying AI and cognitive science in universities around the world before turning his talents to banking. Looking more hippy than hip, he has long, curly hair, crumpled jeans, and John Lennon glasses that are a jarring contrast to the designer-suited bankers who normally work at Hong Kong’s hedge funds. But he is ultimately chasing the same profits.

He’s just doing it by using an artificial intelligence system that can find patterns in the “humungous” volume of market information and data that the human mind “cannot possibly comprehend.”

“Human emotions have certain predictable patterns to them,” he told Quartz. “So it’s amalgamating the predictions of tens of thousands of different predictive patterns that it identifies…and that’s where the AI gets the advantage.”

From back-testing to reality

Hedge funds that use AI to drive their investment decisions have outperformed average industry returns every year for the past seven, except for 2012, according to industry data provider, Eurekahedge.

But it’s a risky industry and the average masks the wide range of returns, with some AI hedge funds making large profits, and others failing spectacularly.

“There are some very disciplined AI programs with strong emphasis on downside protection [a strategy to prevent financial losses] that have been around for more than five years and have delivered double digit returns, and then are others with really volatile returns that the average hedge fund investor would shy away from,” noted Mohammad Hassan, an analyst with Eurekahedge.
This volatility has made itself abundantly clear in each of the last three years, where the overall performance of global equity markets has easily outshone the average hedge fund, AI-driven or not.

Aidyia has conducted extensive testing of its AI system using more than ten years worth of historical data, and CEO Ken Cooper claims it has averaged a very healthy 25% year-on-year return.

Yet historical tests do not always translate into real-world success, noted one senior Wall Street banker, who was not authorized by his firm to speak on the record, but said he would never invest in a hedge fund like Aidyia based on back-tested data alone. “History has never been a good predictor of the future,” he said.

“Money grows in the dark”

Some in the industry, like Gerrit van Wingerden, managing director with Tora, a trading technology group that works extensively with hedge funds and asset managers, believe the success of the AI funds may in fact be underestimated due to the secrecy that surrounds these types of businesses.

“I firmly believe that money grows in the dark and a lot of people who are doing this (AI investing) are keeping their mouths shut as they don’t want people to find out what they are doing and how they are doing it,” he said.

Aidyia’s executives don’t hesitate when asked if AI will eventually replace human traders. “We meet people (in the finance industry) all the time and honestly if I say a robot will be running asset management sometime in the future people go yeah, yeah, I see that future,” said Cooper.

Goertzel nodded in agreement. “You don’t get too much pushback on the idea that AIs will be operating the financial markets in our lifetime,” he said.

Georgia McCafferty is a Hong Kong-based freelance journalist who writes about business, finance, and education.