Quantifying Intuition

MAPPING THE DATA SCIENCE LANDSCAPE

IN THE HEDGE FUND INDUSTRY
Quantifying Intuition
MAPPING THE DATA SCIENCE LANDSCAPE IN THE HEDGE FUND INDUSTRY

Data Science in the Hedge Fund Industry

Machine learning. Big data. Artificial intelligence. Neural networks. The data science tide that has swept across many industries
is increasingly shaping the way hedge funds invest, execute, and run their businesses. Watershed moments seem obvious in
hindsight, but we rarely understand the significance of all the small changes that precede a considerable shift until later.

2016 seems to have marked a tipping point for the hedge fund industry’s mainstream embrace of data science. A number of
firms who would not have cited quant skills or technology as an historic core competency began aggressively exploring and
leveraging data science in their businesses, and we saw ‘quantamental’ become an increasingly popular way to describe firms
incorporating these solutions.

Data science, the umbrella term for the ever-expanding innovations that identify, gather, aggregate and digest enormous amounts
of discrete data points, has permeated the hedge fund industry. Regulators use it to identify bad actors and market irregularities,
as well as gather and aggregate industry information. Investors use it to conduct ongoing operational due diligence on hedge fund
managers, or build internal dashboards to more holistically understand their cross-portfolio risk profile. Counterparties, exchanges
and clearinghouses are using data science to develop new business lines, selling real time data to clients. And hedge fund
managers leverage different solutions to improve their investment approaches, streamline execution and run more efficient and
effective businesses. Even the CFA Institute is close to requiring CFA hopefuls to answer questions on artificial intelligence, big
data and robo-advice in their notoriously rigorous exams.

Quantifying Intuition: Mapping the Data Science Landscape in the Hedge Fund Industry argues
that we have reached a tipping point - data science is now, and will increasingly be, of strategic
importance to all hedge funds in the coming years. This mainstreaming of data science
solutions means managers, investors and counterparties should be informed about and
conversant in the opportunities and challenges offered by these technologies.

This is the first in a series of publications focused specifically on data science at hedge funds. In this framing piece, we focus on
the practical implementation of data science at investment managers. We have also hosted four Artificial Intelligence Summits
across the globe in the last 9 months, and co-sponsored the Learn 2Quant Conference to help discretionary investors better
understand this fast-moving space.

We hope this piece helps explain how investment managers are putting data science theory into practice, and contributes to the
growing body of knowledge about the practical implications of rapidly emerging innovations. We look forward to partnering with
you as you explore the way forward with data science in your organizations, and are happy to answer any questions you may have.

CONTACTS
Shannon Murphy Dan Furstenberg Richard Carbone
+1 (212) 336-1139 +1 (212) 323-3336 +1 (212) 336-7472
[email protected] [email protected] [email protected]

Data Science in the Hedge Fund Industry | June 2017 2

EXECUTIVE SUMMARY
Hedge funds have always differentiated and succeeded by finding an ‘edge’ – leveraging information asymmetries via new sources
of information, or new ways of understanding pre-existing information. We moved from better analysts to channel checks to expert
networks and beyond. Data science is just another evolution - there is more data available now than at any point in history. How
firms effectively leverage this explosion of information is now a strategic business issue that will shape long term success and
sustainability.

Some firms have successfully leveraged data science for more than thirty years; their efforts are not our focus here. Nor does this
piece exhaustively catalogue current available data science solutions. Rather, we explore how hedge funds have more recently
employed and capitalized on the rapidly growing data science industry. We hope to offer practical information to help inform
principals, CIOs and other decision makers to better understand emerging industry trends and how peers may be embedding
these innovations strategically and structurally at their organizations.

There are five core themes of this paper:

1. Data science is an exciting, innovative and critical business strategy issue – but it also remains a completely amorphous one.
This is not limited to hedge funds. As a May 26th Wall Street Journal article reported, CEOs across all industries are feverishly
working to understand the impact of disruptive technologies on their businesses. 1 This is a tough balancing act – these
technologies are increasingly mission critical, but the landscape changes so rapidly, leaders may have a hard time
understanding the landscape, and effectively implementing and adapting over time.

2. Hedge funds are in different stages of embedding data science across their organizations – but funds of all shapes, sizes and
strategies are increasingly dedicating headcount and resources to the space. Jefferies estimates more than 20% of the current
Billion Dollar Club has someone who spends at least half their time focused on the issue, or a dedicated employee managing
data science efforts across the firm. For firms over $8 billion in AuM, it’s more like 40%. Anecdotal information indicates
considerable appetite to dedicate budget to these solutions in the coming years. And a small number of managers report
launching vehicles specifically dedicated to investing in these disruptive technologies as well.

3. “Excel is not a strategy.” PMs, analysts and traders want to understand possible solutions, ranging from unstructured big data
feeds to machine learning tools or technology consulting. Even firms that may not have a long history of quantitative
experience are elevating this issue to understand what potential exists. They explore using data science solutions as a
complement or overlay to longstanding investment processes, better understanding risk profiles and unintended bets, or
improving execution or business processes.

4. Successfully implementing data science across hedge funds can be tricky, and cultural challenges may create unintended
inefficiencies. It’s not enough to say: “Get me a quant, any quant!” Hedge funds are idiosyncratic organizations, and they
need to think strategically about the purpose, implementation and execution of these solutions for their specific firm.
Particularly for those focused on leveraging data science on the investment or alpha-capture side, implementation and
cultural embrace of these solutions becomes paramount to avoid a clash between organizational efficiency and internal
competition.

5. Many questions remain about data science – ranging from legal and regulatory to existential. Will machines eventually out-
invest humans? What is the future for crowd-sourced algo funds? And from a regulatory perspective – what legal issues arise
from these solutions and is there any cross jurisdictional consistency? We’re still in the nascent stages of these innovations,
and their long term implications are still taking shape.

1
“Wall Street to CEOs: Disrupt Your Industry, Or Else” The Wall Street Journal, May 26th, 2017

Data Science in the Hedge Fund Industry | June 2017 3

This paper is broken into four parts. First, because of the diverse and often colloquial nature with which people discuss data
science, we will define the most frequently used terms. We then map the current landscape of data science in the hedge fund
community, from early adopters, to early mainstreamers, to mainstream, to nascent adopters, and briefly discuss each. The third
section broadens the discussion to understand how other actors in the hedge fund industry are leveraging these technologies, and
we explore industry wide trends that are impacting the space. Finally, we offer a vision of what may lie ahead for data science at
hedge funds – exploring opportunities, challenges, and what questions remain unanswered.

A quick exercise in data visualization – the below represents the relative interest in “hedge funds” vs. “data science” as mapped by
Google Trends (which uses machine learning). Since 2004, worldwide interest in hedge funds vastly outstripped that of data science
(peaking in 2008, for obvious reasons), but after inverting in 2014, the two have continued to diverge.

CHART 1
Interest over Time of “Hedge Funds” and “Data Science” from January 1, 2004 to present

Data Science Hedge Funds

100

75

50

25

0
Average Jan 1, 2004 Aug 1, 2008 Mar 1, 2013 Present

Source: Google Trends

Data Science in the Hedge Fund Industry | June 2017 4

I. DATA, DATA EVERYWHERE, AND WAY TOO MUCH TO DRINK
Data, of course, has a nearly limitless history. The earliest recorded exhibit of “data” is thought to be a 20,000-year-old bone (the
Ishango Bone), discovered in Uganda that seems to show Paleolithic tribespeople notching trading activity and supplies. 2 While
we have come a long way from notches and bones – IBM estimates we create 2.5 quintillion bytes of data per day now – basic
issues like how to identify, source, aggregate and understand this information endure. 3

Simply: how do we “get it and digest it?”

Because the answer to that is not simple, and the language around it can often be confusing and pedantic, it's worth taking a
moment to lay out some definitions.

Data science: an umbrella term that captures all methods, processes and systems to extract knowledge or insights from data in its
various forms.

“Big” data: Most of the time this refers to any large set of data or information that can be computationally leveraged to reveal
patterns, signals or trends. Big data has long existed, but the methods for gathering and extracting information from these sets
are new; for example, credit cards have existed for more than 50 years, but investors’ ability to parse credit card transaction data
is more recent. This data may be unstructured or structured, with firms seeking to mine data sets themselves or have both the
data sets and potential signals delivered in a more digestible manner. IBM breaks big data into four dimensions: volume, variety,
velocity and veracity.

Alternative data: very simply, any data arising from sources not considered traditional financial data but that may offer market
insights. Data doesn’t need to be “big” to be alternative, but both alternative and big data are the fuel for data science. Drones
that count cars at Wal-Mart or satellites that track ships are two examples of alternative data sources.

Data visualization: the ability to see and understand patterns or insights from data sets. Tableau, Spotfire and Qlikview are some
commonly used data visualization solutions.

Artificial intelligence (AI): the ability of a computer to do things that used to require “human” intelligence – like speech
recognition, decision making, linguistic translation or vision (image recognition). Two examples: i) Tesla’s continuous updating of
data gathered from its fleet daily to inform its autopilot feature, and ii) Pandora’s system for recommending music.

Machine learning: a way that computers learn from themselves to discover and display patterns in data, adapting to new
information on their own, the way the human brain does. Machine learning takes three forms: i) supervised (more traditional algos
and stats like determining sentiment from having been given positive, negative or neutral tags), ii) unsupervised (interpreting
patterns in a data set automatically – i.e. via “clustering”) or iii) deep (which tries to mimic the work of human neural networks in
the brain to iterate or “learn” when given new information). Siri (Apple) uses machine learning technology to better predict and
understand our natural language questions.

Neural networks: how the human brain makes sense of the world via trial-and-error; retaining information and iterating or
adapting what it “knows” when presented with new data points. Neural networks are the underpinning frameworks for how some
artificial intelligence systems are constructed – namely; how can we make a computer “know” things, then adapt or update this
knowledge when presented with new data.

Natural language processing (NLP): statistical techniques for learning language (i.e. – translation tools). One realm that’s still just
mostly for human intelligence? Sarcasm.

The data market is slated to nearly double by 2020, to upwards of $200 billion. 4 What does all this mean for hedge funds?
Nothing short of a potential sea change in how information is sourced, digested and acted upon.

Markets continuously evolve, but data science and technology have prompted a shift that many feel will transform the entire finance
industry. Traditional data sources have been largely commoditized; data science attempts to solve the new informational asymmetry
question. Those who don’t understand or can’t leverage these opportunities across their businesses may be left behind.

Let’s turn to the hedge fund landscape to understand where managers currently fall across the spectrum of embedding data
science across their organizations.

2
A Brief History of Big Data Everyone Should Read. The World Economic Forum. 25 February, 2015
3
https://www-01.ibm.com/software/data/bigdata/what-is-big-data.html
4
Worldwide Semiannual Big Data and Analytics Spending Guide, International Data Corporation (IDC), October 2015

Data Science in the Hedge Fund Industry | June 2017 5

II. EVERYONE’S A “QUANT” THESE DAYS
Maybe for good reason. Quantitatively focused funds have doubled their assets since 2009 to nearly $1 trillion, and are one of
the fastest growing areas of interest (and allocation) for investors. 5 But you don’t have to be a quant fund to increasingly want to
leverage quantitative resources and personnel in your organization. Firms with various core competencies are actively exploring
how to improve investment theses, risk management or business strategy with the help of different data science solutions.

In the words of one data scientist, “Everyone seems to want to be able to quantify their intuition these days.” An arms race is underway,
with some sourcing various big data solutions and machine learning processes to create better predictive models. It remains to be seen
whether Google searches predict sales of BMWs better than meeting a management team can, or if both are needed?

CHART 2
Hedge Fund Assets Broken Out by Quant and Non-Quant Strategies

3500

3000

2500

2000
Quant Hedge Funds
1500
Non-Quant Hedge Funds
1000

500

0
2013 2016

Source: Hedge Fund Research

Both established managers and new launches are working to figure this out. Jefferies estimates that of the current Billion Dollar
Club – 434 firms at last count of diverse focus, background and location – more than 20% currently have designated headcount
focused on data science, or someone who spends more than half their time on the issue. What those individuals focus on, the
types of teams they lead and their autonomy over budget and spend varies. And of course, these firms have reached a certain
point of maturity to be managing at least $1 billion of AuM.

But new launches are also increasingly hiring data science talent on day one. While it may be too early to call it a trend, we are
increasingly seeing managers launch at inception (or shortly thereafter) with one fundamentally focused analyst and one
quantitative researcher/analyst. What is clear is that a diverse array of hedge funds is increasingly finding it important to
incorporate data science or data scientists into their businesses, at all stages of the hedge fund lifecycle.

A number of managers of all sizes also report they’re starting to field questions from allocators querying: “How do you approach
data?” While open ended in nature and not specifically focused on big data or machine learning specifically, the question clearly
indicates investors understand the changing nature of the information and data landscape, and managers increasingly need to be
prepared to respond. One large investor notes, “We now ask the question, but without presuming an answer. We don’t want
people trying to do things ‘just to do them’ – but we think people should be exploring what’s out there.”

5
Hedge Fund Research, Q1 2017 Industry Report

Data Science in the Hedge Fund Industry | June 2017 6

A review of the hedge fund industry reveals firms currently fall into one of four buckets:

1. Early adopters 2. Early mainstream 3. Mainstream 4. Nascent adopters

CHART 3
Spectrum of Data Landscape Across Hedge Funds
Types of data
Type of firm Headcount Use in firm Budget Overall
science used

Source: Jefferies Prime Services

Hedge funds fall along a spectrum of embedding big data, data aggregation platforms, machine learning solutions, data
visualization tools or technology consulting into their firms.

Early Adopters – there are a number of mature, quant focused firms (typically systematic strategies) for whom data science has
been embedded in their DNA since day one. These shops leverage massive data feeds via API and may have computing power
that rivals the entire U.S. research library system combined. Early adopters have leveraged all that data science has to offer –
artificial intelligence, machine learning, deep learning, big data from countless sources. Their goal is to embed data science
ubiquitously across their firms. They leverage multiple data sets to systematically triangulate new signals, and view the
convergence of these various sets as a useful source of signals, themes and ideas.

They have Chief Data Scientists and teams of individuals dedicated to the space, and use data solutions on the business side,
from HR to compliance or operations. Early adopters are most likely to have started bringing the mountain to Mohammad, by co-
founding a university research center on quantitative finance or data science, or opening outposts in Silicon Valley and staffing
them with engineers from tech firms. These firms are also most likely to have hired high profile employees from places like
Google, Netflix or Microsoft, and constantly hire talent at across all levels to focus on the issue.

From a structural perspective, early adopters are more likely to have discrete verticals of data scientists across the organization,
including, but not limited to: data acquisition, sanitization, digestion and finally application. They are more likely to “pull” data,
testing countless data sets on an endless hunt for obscure or less crowded and more accurate signals – equally focused on what
data sets they’re missing as what they have. For Early Adopters, data is as critical as the PCs on their desktop; as such they have
considerable, dedicated budget for these resources, with appetite to grow.

Early Mainstream – these firms have been working to incorporate data science into their investment process (and sometimes more
broadly across the firm) for the last few years, with the goal of scaling appropriately across the organization. These firms likely
have dedicated or working groups managing data science implementation, and incorporate data scientists into investment idea
generation and research processes. Early Mainstream firms are more likely to have centralized data science groups, who use trial
and error to decide which data sets are most effective, but also may have quant analysts in discrete investment pods.

They may use unstructured data, but more typically leverage multiple structured sets. Early Mainstream firms sometimes overlay
multiple data sets to identify signals, but typically view data through the lens of a pre-existing portfolio theme, as a discretionary
complement “overlay” or for risk purposes. They dedicate budget to data science and understand this allocation is likely to grow
in the coming years.

Data Science in the Hedge Fund Industry | June 2017 7

While Early Mainstream firms have a process for using data science on the investment side, it is less commonly used on the
business side. And they most frequently face a cultural dilemma of balancing their discretionary backgrounds with embedding
these technologies across the firm. There seems to be a trend among some Early Adopters and Early Mainstream in not only
sourcing alternative data, but in allocating venture capital focused on investing in these innovations as well.

Mainstream – firms who have been adding (or looking into adding) data science solutions as a complement or overlay to pre-
existing investment processes. These firms may have thought about adding quantitative personnel on the investment side, or have
started to source structured data to further build out or audit their investment theses. These firms are in the early stages of
understanding how data science may fit into their organizations; mainly seeing data science as a “supplement” to other efforts or
approaches rather than something to be embedded in all aspects of the business. They typically leverage stand-alone data sets,
and their spend is situational, based on particular data sets or technologies.

These firms are most likely to have identified data science as something that can help bolster performance or improve
productivity, but remain in early innings of fully understanding how to effectively incorporate some solutions at their firms.

Nascent Adopters – Nascent adopters may not yet have begun exploring how data science could impact their organizations. They
do not have any headcount focused on data science, nor are there formalized resources dedicated to exploring the space yet, or
solutions deployed across the platform. Many nascent adopters either believe data science is just another fad, or that these
resources won’t serve as an effective business complement.

Even those who aren’t embracing data science yet in their own firms are studying how data science is transforming the industries
they invest in. Energy analysts, for example, may want to understand how Shell leverages big data in its exploration process to
enable its geologists to make more accurate recommendations about where to drill. Big Oil, Big Pharma, tech firms (Google
among those shifting from “mobile first” to “AI first”), retail, and other many industry groups all utilize big data and various data
science solutions now, and analysts increasingly need to incorporate this information in their models.

CHART 4
Data Science Usage Across Various Hedge Fund Verticals

Research Process
Idea Generation Risk Management Execution Business Processes
& Back testing

Identify &
Analyze Assess Execute Streamline
Collect

Big Data Big Data Big Data Big Data Big Data
• Structured or Data Aggregation Data Aggregation • Sourced from
unstructured sets trading venues, Data Visualization
Machine Learning Machine Learning exchanges, and/or
• Social Media, web Data Aggregation
search trends, • Supervised, Data Visualization clearinghouses to
unsupervised or Technology Consulting
consumer optimize execution
reinforced learning Technology Consulting
transactions, and settlement
satellite or drone • Processes reams of
data, governmental data to identify
or regulatory data trends, signals or
outliers
Data Aggregation
• Transforms raw data Data Visualization
to more structured, • Understanding
digestible data sets unintended bets
across the portfolio
Technology Consulting
• Isolating particular
• Assistance in
risks
understanding,
shaping and • Technology Consulting
embedding data
science solutions

Source: Jefferies Prime Services

Data Science in the Hedge Fund Industry | June 2017 8

There is considerable variety in how hedge funds have employed new technologies and solutions at their firms. Firms that try to
incorporate data science without clear purpose or having thought through what’s involved in bringing new innovations on risk
inefficiency, waste and distraction. But when approached thoughtfully and incrementally, data science may improve productivity,
enhance decision making and deliver new sets of actionable data not previously available.

III. DATA SCIENCE IN THE BROADER HEDGE FUND ECOSYSTEM

Counterparties like exchanges and other trading venues are increasingly selling data feeds to clients as new product lines.
Regulators use data science to better understand market activity, and ferret out bad actors or market irregularities. Investors use
data science to build internal risk dashboards to understand comprehensive portfolio risk, as well as conduct ongoing due
diligence on managers. We explore each in turn.

CHART 5
Data Science in the Broader Hedge Fund Ecosystem

• Sources and sells data to clients • Uses machine learning to identify

• Leverage surveillance solutions bad actors and market irregularities
• Builds industry knowledge

Counterparties
and Trading Regulators
Venues

Investors Hedge Funds

• Create proprietary systems that • Leverages across investment

pull all ADVs to gauge changes process, risk management
• Build risk dashboards and business operations

Source: Jefferies Prime Services

Counterparties or exchanges
In part because of shrinking margins, the rising cost of doing business, and lower volumes, counterparties and exchanges are
increasingly turning to selling data or data analysis solutions to augment revenue. The Tabb Group estimates that revenues from
exchange data and technology are up 62% over the last five years, with an overall revenue growth of 16% during the same time. 6

The London Stock Exchange Group recently “raised its earning margin growth forecasts for coming years to future underline its
reorientation from equity trading to high value market data and clearing.” 7 A number of exchanges, including Intercontinental
Exchanges (ICE) and the Chicago Mercantile Exchange (CME), estimate they can grow their data businesses at a rate of about 5-
6% annually for the next few years. 8 And a number of exchanges have made a considerable acquisitive push of data companies in
recent years, converging execution and content – Liquidnet’s May 2017 acquisition of OTAS, as one recent example.

Even large asset managers report continuing to pivot their business models to focus on technology and data. As Blackrock noted
in its 1Q2017 earnings call: “Technology is changing how the world invests and how we interact with clients…[we] intend to

6
The Market Data Deathmatch: The Increasingly Brutal Fight Over Equity Market Data Costs, The Tabb Group, January 2016
7
“LSE Lifts Guidance for Earnings Margin Growth,” The Financial Times, June 12, 2017
8
Surinder Thind, Jefferies Equity Analyst, Asset Managers, Brokers & Exchanges

Data Science in the Hedge Fund Industry | June 2017 9

sharpen the focus on our technology and risk management businesses.” 9 If the world’s largest asset manager believes it’s possible
for considerable forward revenue growth to come from data, software and data analytics – what are the implications for the rest of
financial services?

Allocators & Investors

Allocators and investors are increasingly interested in understanding how their managers are approaching the explosive growth of
available data. They are starting to ask the question of existing and emerging managers, and while firms don’t necessarily need to
have onboarded data science solutions, at a minimum they are starting to have to give thought to the issue.

From a due diligence perspective, data science remains an emerging area – but sophisticated investors are increasingly using
solutions to conduct ongoing due diligence on their managers – in some cases vetting thousands of ADVs to map changes, or
scanning all investor letters for certain keywords or phrases. They’ve adopted these enhanced techniques as part of their overall
risk management processes. A small number of investors are working to build better internal dashboards to understand their
cross-portfolio risk profile and exposures to specific factors by leveraging big data and/or data visualization, often in conjunction
with third party consultants.

Investors have also identified these solutions as worthy investments as an end in themselves. There is at least one fund-of-funds
dedicated to investments in Artificial Intelligence funds, with more than $50 mm invested, and a number of investors report
launching disruptive technology vehicles, with big data, machine learning and AI solutions among their holdings.

Regulators
Regulators leverage data science to: i) better map and understand market activity and ii) identify bad actors or market
irregularities. Scott Bauguess, acting director of the SEC’s Division of Economic Risk and Analysis, remarked in an October 2016
speech, that the agency was using machine learning algorithms to map topic and tonality signals to measure risk::

“Some of these [new] analytical methods are allowing analyses of previously impenetrable information sets –
for example, those without structure, such as freeform text. This has been of particular interest to the SEC,
where registrant filings are often in the form of a narrative disclosure. So, as a result, we have begun a host of
new initiatives that leverage the machine learning approach to behavioral predictions, particularly in the area
of market risk assessment, which includes the identification of potential fraud and misconduct.” 10

The SEC’s MIDAS (the Market Information Data Analytics System) initiative seeks to improve transparency in and understanding
of the market, and the early stage Consolidated Audit Trail (CAT) is working to build a global database of market activity. Both
leverage big data and machine learning technologies. FINRA and the CFTC have also engaged with vendors to develop machine
learning programs for transaction surveillance. The CFTC has launched CFTC 2.0, a part of its LabCFTC initiative, to engage with
new technologies like machine learning, artificial intelligence and other advanced analytics to “improve the effectiveness and
efficiency of the agency in carrying out its day-to-day activities.” 11

These initiatives have helped regulators identify market irregularities or questionable trading patterns by triangulating trading
patterns to uncover insider trading, as in some recent high profile cases. And regulators, exchanges, banks and funds alike
leverage the expertise of firms like Digital Reasoning, whose machine learning technologies can conduct natural language
processing for surveillance of potential internal threats.

Hedge Funds
The various ways hedge funds are embedding data science across their organizations is covered in section II.

9
Blackrock 1Q2017 Earnings Call Transcript, April 19th, 2017
10
Scott Bauguess, U.S. Securities & Exchange Commission “Has Big Data Made Us Lazy,” October 21, 2016 speech before the Midwest Region Meeting of the
American Accounting Association
11
CFTC website: CFTC 2.0, accessed at http://www.cftc.gov/LabCFTC/CFTC2_0/index.htm on June 12, 2017

Data Science in the Hedge Fund Industry | June 2017 10

IV. WHAT LIES AHEAD
Available data continues its exponential growth march – with estimates of 44 zettabytes (or 44 trillion gigabytes) of data points
generated annually by 2020. 12

The most common question from hedge funds that have not begun exploring data science is: “Where do I even start?” We offer
the following as a basic five-step framework for starting to understand and leverage data science at your organization.

CHART 6
Basic Initial Framework to Understand and Incorporate Data Science into Hedge Funds

Source: Jefferies Prime Services

Identify
What do I hope to get out of data science? Why am I even exploring different solutions for my firm?

Improving investment theses. Expanding inputs or ideas. Enhancing risk management. Identifying unintended bets. Running a
more efficient organization. Staying current with industry themes
Is it possible effective implementation of data science could attract incremental investors, new talent, or new enthusiasm about
the research and development (R&D) happening at our firm?

Determine
What expertise can I leverage in better understanding the data science landscape? What resources am I willing to dedicate to
these efforts?

Not everyone needs to hire a dedicated data scientist to work in house. Expertise comes in many forms and may be sourced from
current employees with an interest in the topic (typically a CTO or head of risk), third party consultants, or even experts at
counterparties. Surveying others can help determine the most efficient and effective data science solution for a particular firm
In some cases, an “OQIO” (Outsourced Quant Information Officers) might be of help. Just as Outsourced Chief Investment
Officers (OCIOs) have become a solution for some asset owners to manage their investment portfolios, it’s possible we may
increasingly hear of OQIOs serving as consultants to assist fundamental managers in understanding, selecting and managing
the data solutions that make the most sense for their organizations
Should I attend any events hosted by data science providers or my counterparties that focus on these issues?

12
“The Digital Universe of Opportunities: Rich Data and the Increasing Value of the Internet of Things,” EMC, April 2014

Data Science in the Hedge Fund Industry | June 2017 11

What is my anticipated spend for product, system and implementation? Is it a recurring or one-time cost – and what is my
appetite for that spend growing on a go-forward basis?

Data science solution costs cross a broad spectrum; beyond recurring subscription costs, one time implementation or project costs
can add up. It’s critical to budget base and bull cases for bringing these innovations on board, but costs can scale quite linearly
Where will these allocations live in our budget? Under IT? Under research costs? Do we want to syndicate the costs across
investment teams (if organizationally appropriate)?

Explore
What types of big data solutions exist? How should I be thinking about machine learning down the line – if at all?

There are already an endless – and growing – number of big data brokers. It’s important to understand which will best serve the
goals of your firm and which are least likely to reflect data decay in the short term (i.e. – that the sets get so commoditized,
their signaling benefits are minimized)
Identify the data sets, sectors or themes that are most efficient and effective for your processes
Start to learn about machine learning and other methods of digesting these big data sets for a basic understanding of these
capabilities, even if you don’t incorporate them in the short term

Educate
There is often a cultural component to trying to incorporate data science into our organization

John Maynard Keynes said, “Difficulty lies not so much in developing new ideas as in escaping from old ones.” Principals, CTOs
and risk managers from across the organization need to buy in and fully understand both the opportunities and pitfalls for
bringing new solutions on board. As with most cultural issues, data science needs to be embraced at the top. No group can
advance to efficiently and effectively onboard these solutions without buy in from the principal. Implementation may be
vertical (the principal buys in, and pushes the issue down across the organization), or horizontally shared across the C-suite,
potentially including the CTO, COO, Head of Risk or Trading or senior technology analysts
Sometimes a few training or information sessions are required to fully educate employees and minimize disruption or conflict as
new solutions are brought on

Implement
Roll out solutions on a strategic timeline, allowing sufficient space for incorporating and iterating as needed
Education of these solutions is now an ongoing effort; PMs, analysts or risk managers should work to stay abreast of
emerging innovations

Other Considerations
Do I really need to spend time on this, or is it just another fad?

The exponential growth in data sets and processing solutions has meant that many people don’t even know what they don’t know.
The universe is sufficiently vast and novel that at a minimum, it's worth exploring what might be of use to enhance pre-existing
approaches or improve productivity

Could new data science solutions become part of your marketing or investor relations approach?

We are aware of a number of managers who have started incorporating more quantitative analyses in their pitches, tear sheets and
marketing materials to show how they’re leveraging new data sets or machine learning solutions. They’re expanding the number
of firm employees who meet with potential allocators to include data scientists or quantitative analysts, where appropriate. This
is currently a differentiated approach; but it’s possible 24 months from now, it will be common. There is considerable
opportunity in turning the black box into a grey box and helping inform investors how new quantamental approaches are
supplementing more traditional processes

What are the legal implications of incorporating data science into my firm?

There are numerous legal and regulatory questions that arise for these innovations, including: prohibitions around data scraping
(is your data “clean?”), privacy laws, trespassing, misappropriation, and confidentiality. At this point, there is little clarity
among many of these issues, and little standardization across jurisdictions

Data Science in the Hedge Fund Industry | June 2017 12

How Jefferies Can Help
The scope of the data science universe is enormous, and growing – further complicating things is that there simply is not a one
size fits all approach to adopting these solutions effectively for all firms. The idiosyncratic nature of hedge funds and their
investment processes means that principals, PMs, traders, CTOs and risk managers need to carefully consider the right approach
for their organizations.

Jefferies has centralized its data science expertise, resources and solutions to better help clients approach this issue strategically,
to maximize the return on your invested time covering the issue. Please do not hesitate to let us know how we can be of
assistance as you look to learn more about the ever-changing landscape of data science and its possibilities for your firms, or if
we may follow up on anything discussed herein.

IMPORTANT DISCLAIMER
THIS MESSAGE CONTAINS INSUFFICIENT INFORMATION TO MAKE AN INVESTMENT DECISION.

Please contact your Jefferies representative for copies of the most recent research reports on individual companies.

This is not a product of Jefferies' Research Department, and it should not be regarded as research or a research report. This
material is a product of Jefferies Equity Sales and Trading department, and intended for Institutional Use. Unless otherwise
specifically stated, any views or opinions expressed herein are solely those of the individual author and may differ from the
views and opinions expressed by the Firm's Research Department or other departments or divisions of the Firm and its
affiliates. Clients should assume that this material is not independent of the Firm’s proprietary interests or the author’s
interests. For example: (i) Jefferies may trade for its own account or make markets in the securities referenced in this
communication (and such trading may be entered into in advance of this communication); (ii) Jefferies may engage in
securities transactions that are contrary to or inconsistent with this communication and may have long or short positions in
such securities; and (iii) the author of this communication may have a financial interest in the referenced securities.

The information and any opinions contained herein are as of the date of this material and the Firm does not undertake any
obligation to update them. All market prices, data and other information are not warranted as to the completeness or accuracy
and are subject to change without notice. Past performance is not indicative of future results, and no representation or
warranty, express or implied, is made regarding future performance. The Firm is not providing investment advice through this
material. This material does not take into account individual client circumstances, objectives, or needs and is not intended as
a recommendation to particular clients. Securities, financial instruments, products or strategies mentioned in this material
may not be suitable for all investors. Jefferies does not provide tax advice. As such, any information contained in Equity Sales
and Trading department communications relating to tax matters were neither written nor intended by Jefferies to be used for
tax reporting purposes. Recipients should seek tax advice based on their particular circumstances from an independent tax
advisor. In reaching a determination as to the appropriateness of any proposed transaction or strategy, clients should
undertake a thorough independent review of the legal, regulatory, credit, accounting and economic consequences of such
transaction in relation to their particular circumstances and make their own independent decisions.

OPTIONS ARE NOT SUITABLE FOR ALL INVESTORS. Please ensure that you have read and understand the current options
risk disclosure document before entering into any option transaction. The options disclosure document can be accessed at the
following web address: http://optionsclearing.com/publications/risks/riskchap1.jsp. Please contact Jason Roelke, Head of New
York Derivative Sales for additional information (212-284-2454).

© 2017 Jefferies LLC

Data Science in the Hedge Fund Industry | June 2017 13