You are on page 1of 6

Travail de groupe 

: 5 personnes max

5 pages par personne

On peut rendre le dossier jusqu’à une semaine après l’exam

Representative demand and representative suply

Utility function

Different execution systems to arrange trades

 Quote-driven systems : take or leave : the price is already marked on the product.
The seller quoted the price and the buyer has only 2 choices
 Order-driven systems : it allows you to negociate the price by submitting different
orders. I want to buy something at 500 $, who is going to sell it to me ?
 Brokered trading systems : bnb ect a broker organize the trade
 Hybrid system : mélange des 3

Difference between arbitrage and speculation :

Arbitrage : one-price rule

Limit order and market order à l’exam

Price impact / market impact à revoir

Large spread leads to higher volatility

Relationship between liquidity and volatility

Fundamental value ?

Two strategies
Market structures

4 types

Order-driven
Quoted-driven
Brocker
Hybrid

Omayma Alaoui
Bakkali Saad

Data-driven AI applications to make better informed


lending decisions

Machine learning has had fruitful applications in finance well before the advent of
mobile banking apps, proficient chat bots, or search engines. Given high volume,
accurate historical records, and quantitative nature of the finance world, few
industries are better suited for artificial intelligence. There are more uses cases of
machine learning in finance than ever before, a trend perpetuated by more
accessible computing power and more accessible machine learning tools
 
The term robo-advisor was essentially unheard-of just five years ago, but it is now
commonplace in the financial landscape. The term is misleading and doesn’t involve
robots at all. Rather, robo-advisors are algorithms built to calibrate a financial
portfolio to the goals and risk tolerance of the user. Data-driven AI can be used for
algorithmic trading : Algorithmic systems often making thousands or millions of trades
in a day, hence the term high-frequency trading , which is considered to be a subset
of algorithmic trading. Most hedge funds and financial institutions do not openly
disclose their AI approaches to trading, but it is believed that machine learning and
deep learning are playing an increasingly important role in calibrating trading
decisions in real time. It also can be used in fraud detection : Combine more
accessible computing power, internet becoming more commonly used, and an
increasing amount of valuable company data being stored online, and you have a
perfect storm for data security risk. While previous financial fraud detection systems
depended heavily on complex and robust sets of rules, modern fraud detection goes
beyond following a checklist of risk factors it actively learns and calibrates to new
potential security threats ( It made me think of the movie Minority Report ).

But what interests us here is the use of data-driven AI in lending decisions and
indeed AI can be very useful in the lending process :
 
Underwriting could be described as a perfect job for machine learning in finance, and
indeed there is a great deal of worry in the industry that machines will replace a large
swath of the underwriting positions that exist today

Especially at large companies (big banks and publicly traded insurance firms),
machine learning algorithms can be trained on millions of examples of consumer
data (age, job, marital status, etc…) and financial lending or insurance results (did
this person default, pay back the loan on time, get in a car accident, etc…?).

The underlying trends that can be assessed with algorithms, and continuously
analyzed to detect trends that might influence lending and insuring into the future
(are more and more young people in a certain state getting in car accidents? Are
there increasing rates of default among a specific demographic population over the
last 15 years?).

These results have a tremendous tangible yield for companies but at present are
primarily reserved for larger companies with the resources to hire data scientists and
the massive volumes of past and present data to train their algorithms.

Data-driven AI can also be used in portfolio management for example :


Users enter their goals , age, income, and current financial assets. The advisor then
spreads investments across asset classes and financial instruments in order to reach
the user’s goals.

The system then calibrates to changes in the user’s goals and to real-time changes
in the market, aiming always to find the best fit for the user’s original goals. Robo-
advisors have gained significant traction with millennial consumers who don’t need a
physical advisor to feel comfortable investing, and who are less able to validate the
fees paid to human advisors.

Data-driven AI relies principally on information :


The key to effective decision making is information. The more data you can acquire
and analyze, helping inform and guide the decision, the better the outcome you can
expect.
Would you advocate a spontaneous elopment between veritable strangers,
purchasing a house sight unseen and without an inspection or sending money to a
Nigerian prince who needs help unlocking a fortune in gold?
Of course you wouldn’t, because the best decisions and outcome are rarely based on
limited or misleading information. And yet, for decades lenders have based small-
business lending decisions on a single data point. What’s worsen that data point is
misaligned and often inconsistent with the health and viability of a the business.

Data-Driven Lending
With big data analytics, machine learning and a swelling amount of information at our
disposal, there is no reason why modern lenders should base important decisions on
a single metric.
Non-traditional lenders are leveraging these valuable resources to reinvent the way
small companies manage their cash, attain credit and grow their business. With a
connection to a company-specific data source, like a business bank account, some
non-traditional lenders can now leverage the incredible value of artificial intelligence
(AI) and machine learning risk models to analyze the health of the business, make a
sound lending decision and, in some cases, transfer funds within 24 hours -- all for a
flat transaction fee. In some cases, the underwriting process has become fully
automated and no longer requires personal credit or FICO scores.
It is this intersection of data and AI that is enabling non-traditional lenders to
democratize access to credit and to assist underserved market segments that
traditional lending services have neglected.
Just as retail, hospitality, transportation and other markets have been revolutionized
with advanced technology, so, too, can technological advancement reinvent the
financial industry. But there is no greater need or gap than with small-business
lending.

To help small businesses, we must evaluate them differently. As an industry, we can


do that by leveraging a host of emerging data and computing resources. We must
help small businesses solve their short-term problems and pursue their long-term
opportunities. And, we must create new models that reduce the friction and barriers
of everyday business.

When we do this, we can unlock the full potential of small businesses -- and
everyone wins.

Data-driven IA in a more global strategy of digitalisation :

Today in traditional banks, the average “time to decision” for small business and
corporate lending is between three and five weeks. Average “time to cash” is nearly
three months. Leading banks have embraced the digital-lending revolution, bringing
“time to yes” down to five minutes, and time to cash to less than 24 hours.

That’s the profound result of a top priority for banks around the world: the digital
transformation of end-to-end credit journeys, including the customer experience and
supporting credit processes. Credit is at the heart of most customer relationships,
and digitizing it offers significant advantages to banks and customers alike. For the
bank, successful transformations enhance revenue growth and achieve significant
cost savings.  

As digitization proceeds apace, the dimensions of banks’ digital ambitions vary


among segments and products. Digitization is becoming the norm for retail credit
processes. Personal-loan applications can now be submitted with a few swipes on a
mobile phone, and time to cash can be as short as a few minutes. Mortgage lending
is more complex due to regulatory constraints, yet banks in many developed markets
have managed to digitize large parts of the mortgage journey.

Banks are now treating SME ( small and medium sized companies ) lending as a
digital priority. The reasons are clear: costs are high, and the opportunities to
improve customer experience are significant. Furthermore, both traditional banks and
fintechs already offer compelling digital propositions in SME lending, featuring
dramatically shorter approval and disbursement times—a key factor for customers
when choosing a lender.

Digital is also advancing in corporate lending, though naturally corporate banks are
moving with greater caution and less urgency. Rather than reworking the entire
customer experience, banks are enhancing common processes—for example,
digitizing credit proposal papers and automating annual reviews to improve both time
to yes and “quality of yes.”

Some banks’ digital strategies let corporate-transaction approvers focus their time on
those clients and deals that matter the most. Low-risk credit-line renewals, for
example, can be automated, while valuable human review time is focused on more
complex or riskier deals. And data aggregation can be automated so that relationship
managers (RMs) have the most relevant data and risk-monitoring scores at their
fingertips—including financial performance, industry performance, market and
sentiment data, and pertinent news and external risk factors.

An example of data-driven lending decisions in the automotive industry :

Auto lenders are employing so-called alternative data and, potentially, artificial
intelligence for many purposes :

A key use is approving and disapproving loan applications, which is likely to create
friction with dealers, especially as artificial intelligence catches on. When bigger,
wider sets of data affect the decision on top of traditional measures, most notably
credit history, it can become impossible for a lender to explain precisely why an
individual loan got turned down.

F&I managers may see, for example, customers whose credit scores look acceptable
be rejected for loans as a result of considerations derived from alternative data.

"Would you be comfortable if you don't know how you got there? Are you comfortable
with using that?"These AI systems are not going to give you the direct path, the
coding, to get to that."
There were separate sessions at the AFSA (american financial service association)
conference for alternative data, which is already in widespread use, and for artificial
intelligence, which draws on, and some have referred to as, Big Data. AI is still on the
drawing board for auto lenders. In both sessions, panelists set out to define the
terms.

Alternative data is loosely defined as data that's not traditionally tracked by credit
bureaus, panelists said. They noted, though, that some credit bureaus have begun
gathering alternative data, too. Common examples include credit history for cable TV,
mobile phone and utility bills and salary information.

Anywhere you use a risk score today, you can use alternative data as another
source of information to make decisions. It tells you stability, willingness, ability to
pay. That's really where the power is, leveraging information that is correlated with
credit risk.

What is a FICO Score


A FICO score is a type of credit score created by the Fair Isaac Corporation. Lenders use
borrowers' FICO scores along with other details on borrowers' credit reports to assess credit
risk and determine whether to extend credit. FICO scores take into account various factors in
five areas to determine credit worthiness: payment history, current level of indebtedness,
types of credit used, length of credit history and new credit accounts.

FICO Score

BREAKING DOWN FICO Score


FICO scores range between 300 and 850. In general, scores above 650 indicate a very good
credit history. In contrast, individuals with scores below 620 often find it difficult to obtain
financing at favorable rates. To determine credit worthiness, lenders take a borrower's FICO
score into account but also consider other details such as income, how long the borrower
has been at his job and type of credit requested.

Calculating FICO Scores


To determine credit scores, the Fair Isaac Corporation weighs each category differently for
each individual. However, in general, payment history is 35% of the score, accounts owed is
30%, length of credit history is 15%, new credit is 10% and credit mix is 10%.

You might also like