MH6807

Session 1

Introduction to FinTech
Dr. Jin Song

Aug. 4, 2025
Instructor
• Dr. Jin Song is currently the CEO and Founding Partner of W.I. Consulting, a leading investment
consulting firm in Asia. W.I. Consulting provides high-quality, global investment management and
risk control consulting services mainly for institutional and individual investors in Asia

• Prior to W.I. Consulting, Dr. Jin was

• A Director at Credit Suisse Singapore. Lead the bank Greater China team
• A Director at Pavilion Capital, (a subsidiary of Temasek Holdings); was responsible for Pavilion’s
private equity fund and direct investments in North Asia—China, South Korea, Japan and Taiwan,
• The Head of portfolio management of MCU, OCBC Bank in Singapore
• A senior manager of Deloitte Consulting, USA.; provided private equity advisory service to many
world leading financial institutions and private equity clients in Northeast of U.S.A.
Dr. Jin Song
• Education
CEO and Founding Partner
W.I. Consulting • MBA from Washington University in St. Louis, USA
• Master and a Ph.D. from Nanyang Technological University, Singapore
• Bachelor degree from Xi’an Jiaotong University, China

2
Unique Teaching and Learning Experience
1. Explain basic concepts: main professional terms, industry development trends, etc.
2. Case analysis: Each class will have 1 to 2 actual case studies to deepen students’ understanding
of basic concepts
3. Classroom interaction: through asking questions to stimulate students' independent thinking
ability
4. Class assignments: Except for the first class, students in each class will have a group
assignment,
• Students should combine the theoretical knowledge learned in class
• As a group, complete an assignment and give a lecture in class
5. Lots of reading and thinking
• Read 100-150 pages every day
• What kind of articles to read---professional and unbiased.
• Critical reading-find three questions in each article
3
Course Content
Sessions

Session 1: Introduction to Fintech Session 7: Artificial Intelligence

Session 2: Blockchain Session 8: Machine Learning
Session 3: Cryptocurrency Session 9: Big Data
Session 4: Digital Payment Session 10: Cybersecurity
Session 5: Lending and Digital Banking Session 11: Regulations
Session 6: Class and Quiz (90 minutes) Session 12: Applications
Session 13: Exam

4
Course Schedule (Afternoon Class)

MH6807 Afternoon class 2.00pm - 5.30pm

Session Date Day Time Class Room
1 4/8/2025 Monday 6.30pm - 10.00pm Online
2 11/8/2025 Monday 6.30pm - 10.00pm Online
3 18/8/2025 Monday 6.30pm - 10.00pm Online
4 25/8/2025 Monday 2.00pm - 5.30pm SS-LT29
5 1/9/2025 Monday 2.00pm - 5.30pm NS-LT11
6 8/9/2025 Monday 2.00pm - 5.30pm NS-LT11
7 15/9/2025 Monday 2.00pm - 5.30pm NS-LT11
8 22/9/2025 Monday 2.00pm - 5.30pm NS-LT11
9 29/9/2025 Monday 2.00pm - 5.30pm NS-LT11
10 6/10/2025 Monday 2.00pm - 5.30pm NS-LT11
11 13/10/2025 Monday 2.00pm - 5.30pm NS-LT11
12 16/10/2025 Thursday 2.00pm - 5.30pm SS-LT25
Exam 27/10/2025 Monday 6.30pm - 10.00pm TBD

5
Course Schedule (Evening Class)

MH6807 Evening class 6.30pm - 10.00pm

Session Date Day Time Class Room
1 4/8/2025 Monday 6.30pm - 10.00pm Online
2 11/8/2025 Monday 6.30pm - 10.00pm Online
3 18/8/2025 Monday 6.30pm - 10.00pm Online
4 25/8/2025 Monday 6.30pm - 10.00pm SPMS-LT1
5 1/9/2025 Monday 6.30pm - 10.00pm SPMS-LT1
6 8/9/2025 Monday 6.30pm - 10.00pm SPMS-LT1
7 15/9/2025 Monday 6.30pm - 10.00pm SPMS-LT1
8 22/9/2025 Monday 6.30pm - 10.00pm SPMS-LT1
9 29/9/2025 Monday 6.30pm - 10.00pm SPMS-LT1
10 6/10/2025 Monday 6.30pm - 10.00pm SPMS-LT1
11 13/10/2025 Monday 6.30pm - 10.00pm SPMS-LT1
12 18/10/2025 Saturday 2.00pm - 5.30pm SPMS-LT1
Exam 27/10/2025 Monday 6.30pm - 10.00pm TBD

6
Class Assessment Grading

Class Assessment Grading

S/N Assessment Contents % Individual Group

o
1 Class Participation and Attendance 10 √
2 Group Assessment 20 √ √
3 Quiz 30 √
4 Final Exam 50 √
Total 100

7
Class Assessment
1. Class Participation and Attendance (10%)
• All students are encouraged and expected to actively participate in the class discussion.

• Class Participation points will be given throughout the course based on interactive with the lecturer and
classmates, actively contributing of ideas, probing questions and considered responses

2. Group Assessment (20%)

• I will assign approximately 5 students into each group. All group members are encouraged and expected
to actively participate in the group assignments.

• Group assignment grade will be given based on preparation / readiness, active contribution of ideas,
probing questions and considered responses during class presentation.

• Group assignment presentation detailed instructions, and requirements to be provided in the first class.

8
Class Assessment
3. Quiz (30%)
• A 90 minutes quiz will be conducted after 6 weeks to assess students’ understanding of the topics and
ability to apply the theories and methodology discussed.

• A financial or scientific calculator will be required to answer some of the examination questions.
However, the use of laptops and smart phones will not be allowed.

4. Final Exam (50%)

• A three-hour open book examination will be conducted at the end of this class to assess students’
understanding of the topics and ability to apply the theories and methodology discussed.

• A financial or scientific calculator will be required to answer some of the examination questions.

• The use of laptops will be allowed. However smart phones will not be allowed

9
Introduction to FinTech

10
Fintech Ecosystem & Innovation

11
Technology Disruption: A top down view
Disruption is nothing new, however, the speed and global reach is new

12
Sources: “Global Equity Themes: Disruption – Knocking on everyone’s door”, Credit Suisse, 10 June 2019
Disruption from new technologies: It is nothing new
Change is a constant
• Economic growth as a disruptive
force
• The rise and fall of nations has
been a constant disruptive
development through time
• Sector disruption is a constant

13
Sources: “Global Equity Themes: Disruption – Knocking on everyone’s door”, Credit Suisse, 10 June 2019
Disruption from new technologies: Speed and Reach
Some key developments:
Asset light:
• Most new disruptive business
models appear asset light owing to
the impact of technological
advances.
Speed to market:
• Low capex intensity allows for a
much more rapid deployment of
new services.
• With the help of automation, like AI
and Machine learning
Its reach:
• Today’s technologies can be applied
across multiple sectors and regions
simultaneously

14
Sources: “Global Equity Themes: Disruption – Knocking on everyone’s door”, Credit Suisse, 10 June 2019
Disruption: Technological Innovation
Technology disrupts in more ways
• Potentially most disruptive technologies might impact more broadly than their current universe.
• With an expected time-to-impact of 5-10 years, it seems that little time remains for incumbents to response

15
Sources: “Global Equity Themes: Disruption – Knocking on everyone’s door”, Credit Suisse, 10 June 2019
Technological Disruption at Work: Some examples
Technology allows for new and more efficient services across industries. Technology-induced disruption is only in the
early stages. several examples of how technology is impacting: 1. Banks, 2. Robotics, 3. Asset manager, 4. Retail
Banking Robotics

Retail
Asset Management

16
Sources: “Global Equity Themes: Disruption – Knocking on everyone’s door”, Credit Suisse, 10 June 2019
What is Fintech?
• FinTech-- the use of software and hardware to provide financial services via technological innovation
FinTech companies are often start-ups founded to disrupt financial industry incumbents
• Many incumbents are adopting strategies to respond to the putative threat from start-ups

FinTech impacts in 5 areas

• More cost efficient
• More efficient financial contract
• More effective customer interface
• Enhanced advisory services
• Higher level of security and encryption 17
Sources: “Global banks: Is Fintech a threat of an opportunity”, UBS Global Research, 26 July, 2016
Consumer Perspectives
Mobile payments:
• Usage in emerging markets is 3x the level in developed
• A key challenge for the non-traditional providers was
concern about trust and security

Money transfers:
• Users of money transfers (including FX) by a FinTech
are forecast to rise 73% over the next 12 months
• The largest usage in India (32%) and Kenya (20%) and
the lowest usage in Japan (4%).

Peer-to-peer (P2P) lending:

• Usage in emerging markets is 3x the level in developed
• The threat to banks is real

Robo-advisors:
• Albeit from a lower base, the use of robo-advisors
looks likely to grow exponentially

18
Sources: “Global banks: Is Fintech a threat of an opportunity”, UBS Global Research, 26 July, 2016
Traditional financial services Least likely to be impacted by Fintech
• Investment banking / capital raising
• Payments networks Visa and MasterCard
• Wealth management & financial advisory

19
Sources: “2019 Credit Suisse Fintech Conference Takeaways”, Credit Suisse, 11 June 2019
Bank Response: Partner with FinTech start-ups
Banks that are quick to embrace innovation and digitalisation will be best placed to exploit the benefits of FinTech.
Incumbent banks and FinTech start-ups could potentially leverage off each other's strengths
FinTech partnerships:
Banks offering services in partnership with Fintech companies, and future
• One way that banks have responded to intentions to form partnerships

threats from FinTech competitors has

been via partnerships.
Revenue opportunities:
• The impact of FinTech on revenues are
estimated at 3.8% over the next 3 years
• 1.3% for developed markets
• 5.1% in emerging market

20
Sources: “Global banks: Is Fintech a threat of an opportunity”, UBS Global Research, 26 July, 2016
Regulatory Response
• Regulators overall are positive on FinTech
• MAS and the BoE have both set up "regulatory sandboxes" where financial institutions can experiment
with new technologies in a "regulatory-light" environment
• Even allowing (or expecting) some of these experiments to fail.

21
Sources: “Global banks: Is Fintech a threat of an opportunity”, UBS Global Research, 26 July, 2016
Blockchain /Cryptocurrency
Blockchain /Cryptocurrency

• What is a Bitcoin ? a currency, a security or a commodity

• Why are there so many different cryptocurrencies?
• Do we actually need a Blockchain?
The Blockchain Protocol:
Blockchain is not a single, unified concept; there are different levels of decentralisation from a
permissioned private ledger, such as R3’s Corda , to a fully public, permissionless ledger, such as Ethereum
and Bitcoin.

Centralised De-centralised De-centralised Distributed

24
Sources: “Blockchain 2.0 ”, Credit Suisse, January, 2018
However…. Do you actually need a blockchain?
For a blockchain to be relevant,
you must
1) require a database,
2) need shared write access
3) have unknown writers
whose interests are not
unified
4) not trust a third party to
maintain the integrity of
the data.

Q: Do the big Chinese tech companies use Blockchain on their payment app?
25
Sources: “Blockchain”, Credit Suisse, August, 2016
What is a Bitcoin ? a currency, a security or a commodity
Bitcoin is originally designed as a payment system . However, it never took off as a payment network; only about 1%
of bitcoin transactions are for payments.
Bitcoin is most commonly used for trading. To traders, bitcoin is not about cash, it's about profit.

Bitcoin can’t be a speculative investment and a payment currency at the same time. If it can’t be both, it must be neither

• Security/equity: is secured to the future earnings of a real company, has a claim or liability attached
• Currency: is secured to its government and its tax revenue, has a claim or liability attached
• Commodity: has no obligation or liability to any government, company or other entity

26
Sources: Yahoo Finance
Why is Cryptocurrency So Popular?
Cryptocurrency has become very popular over the past few years as the market caps of leading cryptocurrencies,
such as Bitcoin and Ethereum, have grown tremendously. Key reasons are:

Anonymity: a cryptocurrency transaction remains completely anonymous

Secure: the system is highly secure due to the use of cryptography

No central authority: no central authority that monitors or governs cryptocurrency transactions

Ease of usage: no paperwork required, just the software

No middleman: there is no need of any financial intermediary

27
Sources: “Special Edition Chips & Crypto”, UBS Global research, 26 June 2019
Why There Are So Many Different Cryptocurrencies?

• Since the creation of

Bitcoin in 2009 , more
than 1,500
cryptocurrencies have
been created.
• BTC, ETH, and XRP: the
lion’s share of the
crypto market

Market cap and trailing 30-day average daily trading volume as of Feb. 5, 2018 28
Sources: Coinmarketcap.com; Goldman Sachs Global Investment Research
Payment
Payment

• Why DM doesn’t like Mobile Payment as much as EM does ？

• What is the pain point 3rd part payment company solved？
• How to survive in China 3rd part payment market ？
Infrastructure Set-up Between DM and EM
China has low credit card penetration, but the highest debit card penetration in the world.

Worldwide bank card penetration in 2015 – credit card and debit card
31
Sources: “The Rise of China FinTech Payment: The Ecosystem Gateway”, Goldman Sachs, Equity, Research, August 7, 2017
Digitization of Money
Chinese consumers to leapfrog directly from cash to digital payments, skipping a whole generation of plastic cards.
Number of non-cash transactions in China has increased significantly since 2002

Share of non-cash transactions as a % of total transactions in 2016 (x-axis) vs increase in non-cash transactions from 2002 to 2016 (y-axis) 32
Sources: “The Rise of China FinTech Payment: The Ecosystem Gateway”, Goldman Sachs, Equity, Research, August 7, 2017
From Consumer’s Perspective
• High mobile payment usage in EM than in DM,
why:
• DM trusting their bank more
• DM not trusting the other provider’s security
• What problems/pain mobile payment solved?
None
• Implication:
• Chinese Alipay/Ping An will have more chance
to success expanding in EM than in DM
• Disruptors must built brand and trust
• It takes long time to built
• Can easily be lost

Usage of non-traditional provider services – mobile payments, by country

33
Sources: “Is FinTech a threat or an opportunity?” UBS Global research, July 26, 2016
From Consumer’s Perspective
• Barriers for mobile payments adoption
• Trusting their bank more
• Do not trusting the other provider’s security
• Nothing else, In UK, customers are more likely to
divorce than move their current/checking account
• Convenience
• Cheaper, price sensitivity is low

34
Sources: “Is FinTech a threat or an opportunity?” UBS Global research, July 26, 2016
Current Payment System
Current payment 4- party system: a consumer, a merchant, a card issuing firm and a card network

The breakdown of the merchant service charge

Pain Points
• Antiquated processes
• Data reconciliation
• Lack of automation

35
Sources: “Blockchain”, Credit Suisse, August, 2016
Incumbent’s Response --- Contactless Credit Card
• The popularity of contactless cards, which enable users to tap-to-pay using the same near-field-technology that enables mobile
wallet payments, may boost adoption and awareness of contactless mobile payments.
• Chase, the largest U.S. card issuer, announced it will be giving contactless credit and debits cards to customers who renew
• or open new card accounts in 2019.
• Merchants are becoming more equipped for contactless payments, as 50% of face-to-face Visa transactions occur at
contactless-enabled merchant locations, and 95% of new terminals shipped are tap-to-pay enabled, according to Visa.
• Contactless-payment-enabled terminals and cards are already ubiquitous in Europe, facilitating the region's rapid adoption of
mobile wallets that use the same tap-to-pay technology.

Contactless Payment-Enabled Merchants Growing 36

Sources: “Apple Looks to Financial Services for Device Loyalty”, Bloomberg Intelligence, Nov. 29, 2018
Blockchain applications: Cross Border Payment—Low hanging fruits
Bank-to-bank fund transfer, especially the cross-border payment presents some of the lowest-hanging
fruit for blockchain disruption.

• These systems, such as SWIFT, are decades old, have very limited flexibility and face growing security threats.
• They are also slow and costly – with cross-border wire-payments taking days to clear with fees as high as 10%.
• Movements of funds between correspondent accounts in the same country or same monetary zone can happen
through local clearing systems. SWIFT is not mandatory even if it might be used as well.
37
Sources: “Blockchain”, Credit Suisse, August, 2016
From Consumer’s Perspective
• Cross border money transfer and FX services
• There is a considerable delays
• Significant fees and commission attached
• More emerging markets (EM) consumers use
non-traditional cross border money transfer
systems than do so in the developed markets
(DM)

Cross border money transfer--- current and likely future use

38
Sources: “Is FinTech a threat or an opportunity?” UBS Global research, July 26, 2016
Lending/Digital Banking
Lending/Digital Banking

• Who wants to be a Bank?

• What are the biggest challenges if you are not a High Tech company
and wants to be a digital bank?
• Why the incumbent banks are not afraid of Digital Banking
challenges?
Who wants to be a Bank?
Bank share prices have destroyed significant value over the last 10 year
• If you invest 100 Euro in the European banks ten years ago, it would worth 83 Euro today, including dividends
• If you invest USD$100 in US high tech firms ten years ago, it would worth USD $445 today, including dividends
The reasons are
• Low rates and higher capital requirements are major culprits of pressure on profitability NDXT: Nasdaq 100 Technology sector index
SX7P: Stoxx Europe 600 Banks Index
• Compounded by vast settlement costs for litigation and customer refunds

41
Sources: “Big banks and the bigtech, fintech & digibank incursion. What is at stake?”, UBS Global research, June 18, 2019
What are Big Tech Firms thinking?
Why high margin, high P/E ratio big tech firms want to get into low valuation banking business?
Why would a big tech company wish to gain exposure to the regulation, balance sheet leveraged and risks
attaching to banking?

42
Sources: “Big banks and the bigtech, fintech & digibank incursion. What is at stake?”, UBS Global research, June 18, 2019
The Answer is Trust
The banking services, lending customer money is an obvious way to extend the core business which is usually
the sale of some product or service, eg. Finance provided by car companies
Lending, rather than storing or investing, generally carries much lower reputations risk

• About a third of auto manufacturer

EBITDA is generated by lending
operations
• Finance is a core component of the
business model (% of 2018 EBITDA)

43
Sources: “Big banks and the bigtech, fintech & digibank incursion. What is at stake?”, UBS Global research, June 18, 2019
Lending Club
• Perhaps one of the most significant changes in the U.S. lending landscape post the 2008-10 Financial Crisis
has been the emergence of marketplace lenders, also called peer-to-peer (P2P) lenders.
• The original thesis behind the marketplace lenders model was that it would disrupt legacy bank consumer
loan products – primarily credit cards – resulting in significant consolidation of banks' card receivables into
Marketplace lenders’ primary product, amortizing personal loans, which would then be sold to a wide range
of investors, from credit hedge funds to individuals.

44
Sources: “Big banks and the bigtech, fintech & digibank incursion. What is at stake?”, UBS Global research, June 18, 2019; company website
Lending Club: However, the assumptions were challenged
• Despite its online-only presence, its cost base was still outsized
relative to its origination volume. So, while it has been able to offer
lower interest rates as an inducement to consumers, it has also had to
undertake a significant reduction in its operating costs and platform

• Its marketing and acquisition model ultimately proved similar to that

of the card lenders. Notwithstanding its digital tools, Lending Club's
most successful channel for origination is direct mail, an area where it
does not have a particular cost or targeting advantage.

• Its underwriting model hasn't produced superior risk-adjusted

returns. Through 2016-17, the company had to adjust its
underwriting, raise pricing, and largely exit the non-prime segment,
where losses were ahead of internal forecasts, and focus on the more
competitive near-prime and prime segments.

45
Sources: “Big banks and the bigtech, fintech & digibank incursion. What is at stake?”, UBS Global research, June 18, 2019
Funding Cost is the Biggest Challenges for a Digital Bank
Digital challenger banks paying up to 120bps more than incumbents for one year deposits

• Getting customers to move in the short term

in developed markets requires newcomers
to pay up for customer deposits in particular

• The gap for one year deposits between

digital challenger and incumbent banks in
the UK deposit market is up to 140bps

UK digital banks pay up to 140bps more than incumbent

banks for one year deposit money

46
Sources: “Global banks: Is Fintech a threat of an opportunity”, UBS Global Research, 26 July, 2016
Bank Perspectives: Payment and FX Remittance and P2P are largest threats
How strongly are retail Fintech services perceived as a threat to banks –developed vs. emerging
• Payments: Banks consider the major threat from FinTech to come from the payment businesses
• FX, remittance & P2P: Disintermediation risk from FinTech was considered to be next highest in the area
of FX and remittance

47
Sources: “Global banks: Is Fintech a threat of an opportunity”, UBS Global Research, 26 July, 2016
Artificial Intelligence Machine Learning and Big Data
Artificial Intelligence
• Machine (computer)
+ Mathematics (algorithm) Machine Learning
= Human intelligence
• Understanding Human • Enables computer systems to Big Data
Intelligence improve with experience and
• Use advance AI technologies to data. • Large in volume, velocity and
re-create human intelligence • Representing the world as a variability
nested hierarchy of concepts • Generated by individuals
• Achieves great power and • Generated by business processes
flexibility • Generated by sensors

Artificial Machine Big

Intelligence Learning Data

48
Sources: “If Data is the New Oil, is AI the New Internal Combustion Engine?”, UBS Global Research Q Series, April 4, 2018
What is Artificial Intelligence?
Using Machine (computer) + Mathematics (algorithm) == To Recreate Human Intelligence
• Understanding Human Intelligence
• Use advance AI technologies to recreate human intelligence

49
Sources: “If Data is the New Oil, is AI the New Internal Combustion Engine?”, UBS Global Research Q Series, April 4, 2018
Where are we now at AI?
• Among categories of human intelligence, most AI efforts
today have focused on perception as it is the easiest to
model into algorithms (e.g. the "lowest hanging fruit") with
Deep Learning being the best algorithm.

• There is also an significant amount of focus on planning (e.g.

decision making, a critical element for autonomous vehicles)
and language (e.g. human communication) though these are
both farther behind perception.

• The bottom line is that there we are incredibly early in Missing

mapping true human intelligence and the vast majority of AI
the landscape is un-explored white space.

50
Sources: “If Data is the New Oil, is AI the New Internal Combustion Engine?”, UBS Global Research Q Series, April 4, 2018
Deep Learning Adoption: It Is Still Early
Adoption of deep learning technology is still in the "early market" and may not cross the chasm, where the vast
majority of profits are made, within the next several years

1. Immature AI Technologies, as demonstrated by

the preponderance R&D initiatives over
production-ready deployments
2. A lack of system integrators to help mainstream
customers extract value from AI technologies
3. A lack of ROI frameworks to help managers
interpret the risk and rewards of their
investments in AI technologies

The technology adoption lifecycle for deep learning

51
Sources: “If Data is the New Oil, is AI the New Internal Combustion Engine?”, UBS Global Research Q Series, April 4, 2018
Still Far From Human Level Intelligence
Still Far From Human Level Intelligence
• Industrial successes mostly based on supervised learning requiring lots of human-labeled data implicitly
defining the relevant high-level abstractions
• Learning relatively superficial clues, sometimes not generalizing well outside of training contexts, easy to
fool trained networks
Humans outperform machines at unsupervised learning
• Humans are very good at unsupervised learning, e.g. a 2-year old knows intuitive physics
• Babies construct an approximate but sufficiently reliable model of physics.
• How do they manage that?
• Babies interact with the world, not just observe it.

52
Sources: “Artificial Intelligence Hits the Barrier of Meaning”, By Melanie Mitchell, The New York Times, November 6, 2018
What is Machine Learning?
Machine Learning can be seen as a model-independent (or statistical or data-driven) way for recognizing
patterns in large data sets.

ML has two phases:

1. Training
2. Testing

53
Sources: “Big Data and AI Strategies Machine Learning and Alternative Data Approach to Investing” JP Morgan, May 2017
History of Machine Learning
• Deep learning has had a long and rich history, but has gone by many names, reflecting different philosophical viewpoints
• Deep learning has become more useful as the amount of available training data has increased.
• Deep learning models have grown in size over time as computer infrastructure (both hardware and software) for deep learning has improved.
• Deep learning has solved increasingly complicated applications with increasing accuracy over time

1990s multi-layer neural

1945 The standard 1960s Widrow-Hoff 1980s Webos (1970) and network was not only
McCulloch-Pitts refined the learning Rumelhart (1986) time-consuming, but also
model of neuron rule created newer models error-prone as the number of
hidden layers increased

1st AI Winter 2nd AI Winter

1958 Franklin
Rosenblatt built the 1969 MIT theorist Marvin 1989 US Postal Service 2006 Geoffrey Hinton
first modern neural Minsky wrote a scathing Hand Written project demonstrated that a
network critique in his book titled multi-layer neural network
model---the “Perceptrons” could be efficiently trained
perceptron

54
Sources: Goodfellow I., Y. Bengio, A. Courville. (2016). Deep learning. MIT Press. www.deeplearningbook.org.
History of Machine Learning: 1945
A simple calculating model for a neuron was suggested in 1945 by McCulloch and Pitts which could compute a
weighted average of the input, and then returned one, if the average was above a threshold and zero,
otherwise.

The standard McCulloch-Pitts

model of neuron

55
Sources: “Big Data and AI Strategies Machine Learning and Alternative Data Approach to Investing” JP Morgan, May 2017
History of Machine Learning: US Postal Service Hand Written project
1989
• The first practical application of neural
networks to massive data sets
• AT&T Bell Labs used data from the US Postal
Service to decipher hand-written zip code
information

MNIST handwritten digits database

56
Sources: “Big Data and AI Strategies Machine Learning and Alternative Data Approach to Investing” JP Morgan, May 2017
How Did Computer Learn To Recognize Hand Written Digits (0)
MNIST handwritten digits database
28

28 28 x 28 = 784

Each neuron holds a number between 0-1

www.youtube.com/watch?v=aircAruvnKk
0.30 0.00 1.00
57
Sources: “Neural Network”, www.3blue1brown.com
How Did Computer Learn To Recognize Hand Written Digits? (2)
Weight and biases
Number of weights: 784x16+16x16+16x10
Number of biases: 16+16+10
Total No. of Variables: 13,002

– w is weights
is the sigmoid function – b is bias

What does machine learning mean?

To Find the right weights and bias

58
Sources: “Neural Network”, www.3blue1brown.com
Does Deep Learning Actually Learn?
Deep learning doesn’t actually “learn”. It is only optimization.
• Deep learning doesn’t actually learn in the traditional sense of the word.
• Deep learning’s approach fails to add up to actual “learning”, because
• Back-propagation doesn’t correlate to the brain
• Deep learning more as a kind of optimization than the truly intelligent computation we find in the brain
• Deep learning requires too much training and too reliant on data with significant bottlenecks in what it is capable of doing
• Deep learning is nothing but an optimization process of machine learning that requires constant supervision and training periods
for highly selective tasks
• Until 2015, all ad targeting algorithms across the industry were running on mere logistic regression. In fact, that’s still true to a
large extent today
• It is the reason why so many of the ads you see online are desperately irrelevant.
• The social media bots used by hostile state actors to sway public opinion have little to no AI in them. They’re all extremely
primitive
• True artificial intelligence still has a long ways to go
59
Sources: “Does Deep Learning Actually Learn? AI still has a long ways to go”, Michael K. Spencer , Mar 18 2019, www.medium.com
Factors Lead to Big Data Revolution

Three trends that enabled the Big Data revolution:

1. Exponential increase in amount of data available
2. Increase in computing power and data storage
capacity, at reduced cost
3. Advancement in Machine Learning methods to
analyze complex datasets

Artificial Intelligence
A broader scheme of enabling
machines with human-like
cognitive intelligence 60
Sources: “Big Data and AI Strategies Machine Learning and Alternative Data Approach to Investing” JP Morgan, May 2017
Alternative Data
At the core of the Big Data transformation are the new sources of data that can provide informational advantages.
The advantage given by data can be in the form of uncovering new information not contained in traditional sources,
or uncovering the same information but at an earlier time.

61
Sources: “Big Data and AI Strategies Machine Learning and Alternative Data Approach to Investing” JP Morgan, May 2017
How Target Figured Out a Teen Girl was Pregnant Before her Father Did?

Forbes
February 16, 2012

62
Sources: “How Target Figured Out A Teen Girl Was Pregnant Before Her Father Did”, Kashmir Hill, Forbes, February 16, 2012
Attributes of an Alternative Big Data Set
Attributes of datasets that are directly relevant for investment professionals such as mapping datasets to an asset class or investment
style, alpha content, quality of data, technical specifications, etc

63
Sources: “Big Data and AI Strategies Machine Learning and Alternative Data Approach to Investing” JP Morgan, May 2017
Information Content of an Alternative Big Data Set
• Alpha content has to be analyzed in the context of the price
to purchase and implement the dataset.
• Costs of alternative datasets vary widely
• Trading strategies based on alternative data are tested, and
Alpha is estimated from a backtest.
• These tests can find whether a dataset has enough alpha to
make it a viable standalone trading strategy. These
situations are rare.
• Most data have a small positive Sharpe ratio that is not
sufficiently high for a standalone investment strategy.
Despite this, these datasets are very valuable, as the signals
can be combined with other signals to yield a viable
portfolio level strategy.
• Investors should not be surprised to come across
alternative datasets with no alpha content.

64
Sources: “Big Data and AI Strategies Machine Learning and Alternative Data Approach to Investing” JP Morgan, May 2017
Who Are Those Guys: Black Hats, White Hats and Grey Hats
“If you know the enemy and know yourself, you need not fear the result of a hundred battles.
If you know neither the enemy nor yourself, you will succumb in every battle”
<< The Art of War>> Sun Tzu

A Look Into The Hacker Landscape

• Hackers play a central role in the IT security landscape, are a diverse population, and engage in hacking
for a variety of reasons.
• Some hack out of criminal motives; some hack to satisfy their egos or gain peer recognition; and others
hack out of curiosity and for intellectual stimulation.
• Some hack alone, and some hack in groups, and they do so from all over the world. Interestingly, there
appears to be some regional roots in hacking preferences—with
• Vietnamese often targeting ecommerce,
• Eastern Europeans predominantly attacking financial institutions,
• Chinese focusing on intellectual Property.
65
Sources: “Cybersecurity: The Cloud has no walls”, Credit Suisse Equity Research, Sept. 5, 2017
Who Are Those Guys: Black Hats, White Hats and Grey Hats
“If you know the enemy and know yourself, you need not fear the result of a hundred battles.
If you know neither the enemy nor yourself, you will succumb in every battle”
<< The Art of War>> Sun Tzu

A Look Into The Hacker Landscape

• Hackers play a central role in the IT security landscape, are a diverse population, and engage in hacking
for a variety of reasons.
• Some hack out of criminal motives; some hack to satisfy their egos or gain peer recognition; and others
hack out of curiosity and for intellectual stimulation.
• Some hack alone, and some hack in groups, and they do so from all over the world. Interestingly, there
appears to be some regional roots in hacking preferences—with
• Vietnamese often targeting ecommerce,
• Eastern Europeans predominantly attacking financial institutions,
• Chinese focusing on intellectual Property.
66
Sources: “Cybersecurity: The Cloud has no walls”, Credit Suisse Equity Research, Sept. 5, 2017
Traditional Methods Of IT Delivery Are Internally Oriented
Traditional Methods of IT Delivery Have an Easily Identifiable, and Therefore Defensible, Perimeter

Traditional methods of IT : Firewall

• Traditional methods of IT delivery are
internally oriented.
• Applications run on an internal datacenter
and are accessed by a demarcated
sedentary workforce.
• The nature of this architecture lends itself
to a walled garden defense; build tall
strong walls at the perimeter and defend
those walls carefully.

67
Sources: “Cybersecurity: The Cloud has no walls”, Credit Suisse Equity Research, Sept. 5, 2017
The Cloud Has No Wall
The Perimeter Becomes More Difficult to Define in a Hybrid Cloud World

• The cloud, being by its very nature

distributed, starts to dissolve the concept
of network perimeter.
• The traditional corporate network has a
clearly defined, and therefore defensible,
perimeter.
• As workloads move out of data centers to
cloud environments, it becomes less clear
at what point the bounds of the
corporate network begin, and where they
end

68
Sources: “Cybersecurity: The Cloud has no walls”, Credit Suisse Equity Research, Sept. 5, 2017
How Do Cybersecurity Companies Adapt?
As compute becomes increasingly distributed, beyond the traditional network, security must prioritize breaking
down data silos and adopting an identity-centric model (for both users and machines).

Analytics as well as automation/orchestration become paramount

in this model, where data from various security controls must
inform security operations as a whole and facilitate speedier
response.
Ultimately, several sub-sectors stand to benefit, including
• Next- Generation Endpoint Security Analytics
• Security Orchestration, Automation and Response
• Security-as-a-Service
• Next-Generation Identity and Access Management
• Encryption and Application Security

69
Sources: “Cybersecurity: Investor Guide to RSA 2019”, Credit Suisse Equity Research, March 1, 2019
Shift From Prevention To Detection To Response
All large organisations to assume that they are being attacked. If you cannot keep all the attacks out you must
assume that you need to plan to be resilient in the face of them existing

• It is increasingly accepted by practitioners that information security isn’t capable of predicting or

preventing targeted attacks, particularly in the context of an innovating attacker and a mercurial attack
surface.
• In countering the increasing speed, sophistication, scale, and variety of attacks, companies are more
likely to invest in enhancing detection and response capabilities, as well as data origination and endpoint
monitoring as opposed to prevention-centric security measures.

70
Sources: “Cybersecurity: The Cloud has no walls”, Credit Suisse Equity Research, Sept. 5, 2017
Next-Generation Endpoint Security Analytics—Palo Alto Networks
US cybersecurity company Palo Alto Networks (NYSE: PANW) acquired Israeli cybersecurity company Secdo

What does Secdo do?

• Combines integrated endpoint detection and response (EDR) with security
automation.
• Provides thread level visibility by deploying sensors across all endpoints
• Automatically investigates and responds to alerts from any third-party
solution
What does Palo Alto Networks’ Traps do?
• Provides endpoint protection
Once Secdo’s advanced EDR and data collection and visualization capabilities
integrated with Traps, the combined capabilities will
• Provides a new source of rich data, which will feed into the Logging Service.
• Provides customers the capabilities they need to swiftly and accurately
detect, respond to, and even stop cyberattacks across cloud, endpoint, and
the network
71
Sources: “Cybersecurity: The Cloud has no walls”, Credit Suisse Equity Research, Sept. 5, 2017
What Should Financial Regulators Do?
FinTech is fundamentally transforming the financial industry
•The smartphone is becoming our bank.
•Computer programmes are offering investment advice.
•People can now lend directly to one another online.
Unregulated FinTech companies are now providing many financial services
•They will challenge traditional financial institutions
•disrupt and even displace some of them but will not replace them
Financial institutions are not sitting still
•They are actively harnessing technology to enhance product offerings and improve efficiency.
•They are setting up in-house FinTech units to replicate the start-up mentality.
But the FinTech phenomenon is not without risk
•The rapid pace of innovation and adoption throw up many questions for the industry and financial regulators
72
Sources: "FinTech – Harnessing its Power, Managing its Risks“, Mr Ravi Menon, Managing Director, MAS, 2 April 2016
Global Fintech Themes
1. Payment and Digital Banking 2. Bank Technology and Infrastructure:
• Payment as an Entry Point • Outsourcing or In-House
• Cryptocurrency /Blockchain • Banking-As-A- Service
• CBDC • Open Banking API
• Buy Now, Pay Later
• Unbanked and Underbanked

3. Distribution Channel of Financial Services 4. A Long Path to Profitability

• Embedded Financial Services • Fintech is Local/Regional
• SMB Financial Services Under One Roof • Fragmented Industry
• Wealth and Asset Management • A Long Path to Profitability

73
Sources: “Global Fintech”, Credit Suisse Equity Research, May 2021
Fintech Applications

1. InsuTech

2. RegTech
3. LegTech
4. Real Estate
5. Wealth Management

74
 Readings:
Key Readings:
1. “Fintechs: A new paradigm of growth”; Mckinsey & Company, October 24, 2023 | Report www.mckinsey.com
2. “Cambridge Digital Mining Industry Report”; University of Cambridge, Judge Business School, April 2025,
https://www.jbs.cam.ac.uk/wp-content/uploads/2025/04/2025-04-cambridge-digital-mining-industry-report.pdf
3. “How the data mining of failure could teach us the secrets of success”, available at
https://www.technologyreview.com/s/613229/data-mining-reveals-the-hidden-secret-of-human-failure-and-how-to-turn-it-into-success/
4. “1 in 3 bank jobs will merge or change in next few years: IBF, MAS report”, The Business Times, 24 April, 2019
5. “2019 Credit Suisse Fintech Conference Takeaways”, Credit Suisse, 11 June 2019
6. “Global banks: Is Fintech a threat of an opportunity”, UBS Global Research, 26 July, 2016
7. “Global Equity Themes: Disruption – Knocking on everyone’s door”, Credit Suisse, 10 June 2019
8. “ Global Fintech Report Q3 2018” CB Insights, available at https://www.cbinsights.com/research/report/fintech-trends-q3-2018/
9. Citi GPS, 2016, Digital Disruption: “How FinTech Is Forcing Banking to a Tipping Point”, available at
https://ir.citi.com/SEBhgbdvxes95HWZMmFbjGiU%2FydQ9kbvEbHIruHR%2Fl e%2F2Wza4cRvOQUNX8GBWVsV
10. “The Fight for the Customer McKinsey Global Banking Annual Review 2015”, McKinsy&Company, available at
https://www.mckinsey.com/industries/financial-services/our-insights/the-fight-for-the-customer-mckinsey-global-banking-annual-review-2015

75
Readings:
1. “Where are incumbents under the greatest threat from AI?”, UBS Global Research Q Series, December 3, 2018
2. “Artificial Intelligence Hits the Barrier of Meaning”, By Melanie Mitchell, The New York Times, November 6, 2018
3. “If ‘Data is the New Oil’ is AI the New Internal Combustion Engine?”, UBS Global Research Q Series, April 4, 2018
4. <<Deep learning>> MIT Press, GoodfellowIan, YoshuaBengio, Aaron Courville. (2016).
5. “This High-Tech Solution to Disaster Response May Be Too Good to Be True”, Sheri Fink, The New York Times, August 9, 2019
6. “Machine Learning In Finance Limits and Potential”, UBS Global Research, June, 2018
7. “Introduction to Deep Learning”, UBS, Quantitative Monographs, October 25, 2017
8. “Big Data and AI Strategies Machine Learning and Alternative Data Approach to Investing” JP Morgan, May 2017
9. Cybersecurity and Machine Learning with Stuart McClure, Cylance” Stuart McClure and Michael Krigsman, January 18, 2019,
available at www.cxotalk.com
10. “Industry snapshot; GIS000192 Cybersecurity Industry”, AcquisdataGlobal Industry SnapShot, July 29, 2019
11. “Cybersecurity: The Cloud has no walls”, Credit Suisse Equity Research, Sept. 5, 2017
12. "FinTech –Harnessing its Power, Managing its Risks“, Mr Ravi Menon, Managing Director, MAS, 2 April 2016
13. “Fintech Sector Strategy: Securing the Future of UK Fintech”, UK HM Trearsury, March, 2018
14. “EU action plan fintech factsheet”, European Commission, August, 2018
15. “Fintech Action Plan: for a more competitive and innovative European Financial sector”, European Commission, August, 2018 76