Our primary goal is to help you to understand FinTech and to become more confident and persuasive in your ability to analyze and make recommendations to executives within the finance industry regarding how to react to these changes. This FIRST MOOC ON FINTECH IN ASIA-PACIFIC offered by HKUST presents the views of several professors from the top business school in Asia as well as perspectives from industry professionals. HKUST has been ranked for many years as the number one EMBA program in the world, as the number one Finance program in Asia, and as the top MBA program in Asia by multiple independent rating and review journals and surveys. This course offers the combined strengths of HKUST in Business, Finance, and Technology as one the world's leading academic institutions on an important area of technology and business innovations. For students from outside Asia, this also gives an insight into changes happening in the most advanced FinTech markets of the world, since Asia leads worldwide in FinTech adoption and creative innovations. This course (FinTech Foundations and Overview) is part of the HKUST FinTech specialization, there will be 3 other courses in the specialization including FinTech Security & Regulation (RegTech), FinTech Risk Management, FinTech Innovative Disruption: Implications for Society.
3.6: Big Data and Analytics (Part 1)
Loading...
From the course by The Hong Kong University of Science and Technology
FinTech Foundations and Overview
23 ratings
From the lesson
The Tech of FinTech
The Technology enabling and supporting FinTech is important to understand, and this module provides a brief introduction to each of the areas of technology that enable FinTech business changes, including identity and privacy technologies, blockchain and encryption, big data analytics, AI and automation, and consumer tech innovations. This introduction is designed to be at a level appropriate for business managers or students of business, and is not intended to excite or engage tech geeks.
Meet the Instructors
Theodore Henry King CLARKAssociate Professor
Information Systems, Business Statistics and Operations Management Department
Welcome back to our course on FinTech foundations and overview.
In this session on FinTech technologies,
we're going to look at the issue of big data and data analytics.
And we're going to take a look at what does this mean
both in terms of the opportunity and how
this technology or capability enables FinTech applications to develop.
With me I have Dr Hilton Chan,
adjunct Professor of HK university and CEO of his own company and the FinTech space.
So, as we talk about big data,
how important is data,
data mining, data analytics to fintech?
Yes, I mean in the old days you know,
a lot of the data residing with the government or some of the big companies,
they have control of all the data like the banks,
the telecommunication companies et cetera.
But now with Internet,
a lot of the data are being uploaded onto the Internet,
including news, a lot of articles we read about a person, blocks you know.
There comes a wealth of information,
wealth of knowledge that is out there.
And sometimes the data is with these Internet companies like for example;
Alibaba knows a lot about you,
uber knows a lot about you.
Airbnb may know a lot about you,
and so they may have the data and other databases
that the big banks or telecoms or insurance companies don't have.
Yes, that's right. I mean nowadays everybody will use search engines.
Obviously, you leave down a digital trail on a lot of the search engines.
Before I want to visit Australia,
I probably would key in those information,
and you know as for what are the scenery sports in Australia, hotels, et cetera.
So, that is kind of so-called for knowledge before I actually do the planning then
I'll really release some of my quoted data on the Internet, on the search engine.
So the search engine will be able to pick it up,
and know that how many people in a summer are planning to come to Australia.
So, those information are not even available to
the government's visa office in the old days.
How does that help FinTech?
Yes, for example, if those information are available,
then you will consider some of the stops you know.
So, whether those informations will affect
the tourist industry in Australia or whether let's say,
if there is a certain disaster in a particular country,
would that have any impact upon the tourism,
impact upon certain stocks in the market?
Maybe, impact on the currency valuation of the country's currency.
That will be as well. Yes.
Yes.
And because we know that in some of the countries that'll mean.
Tourism is big.
Tourism, foreign students, immigrants and
then all these are things that they all have impact upon the currency as well, yes.
We may be able to gauge public sentiment based on data mining as well.
Yes.
Okay. So, that can matter but one of the challenges of data mining
is that data is often of poor quality.
Yes.
So, what do we do about this problem of data quality
or unclean data if you will or chaotic data?
Right. Now very often that on the Internet
right now there is what we consider as not only factual data.
There is a lot of opinions,
a lot of rumors and a lot of kind of quote and quote.
And then you don't even know what is the original source of that.
And then also, stated bill point at other data,
and then once you find out the original source,
the story could be totally different et cetera.
Now these are unavoidable.
So lots of fake news.
Lots of those information being unconfirmed and that are happening.
Right.
Of course there's the good and the bad part of it.
I mean, the exciting part is for technology people is that,
how could you design a search engine or design some sort of engine to analyze
the ocean of data to identify some of the good ones or
the quantum ones that fits your problem solving requirement.
So, we have more and more data,
but that data may not be easy to use?
Yes, and that's why it comes up with a lot of data analytics models,
a lot of engines and people apply statistics
on different models to try to find out, for example,
I mean one something very simple basic you may say
if different sorts of data are pointing to that direction,
we consider that a piece of data will be something closer to factual.
But on the other,
when people understand that and people trying to create a so-called quoted,
a fake fax from different source so that it would see the engine, right?
Right.
So, it's not always the good and evil is always playing against each other in real life.
Some people argue that the meltdown in
2010 financial markets with nine percent drop in one day was due to
some attempts to manipulate the market with
false signals or information that caused algo-trading to go around.
Yes, I mean some say it is, some say it's not.
But obviously, we will realize that the program training having been here,
it's almost like somebody hurts us, you know.
You get the wrong signal.
Yes, and say a fire alarm,
an alarm bell is on whatever,
it just like screaming and everybody panics and then everybody react to the market.
And also, research discovered that especially in sentiment analysis is,
negative news caused people to react faster than positive news.
So, it seems that people are more paranoid than
being optimistic in making decisions on certain things.
Now, if we know that we can design that into
our models and we can say that's part of our systems.
Yes.
Part of our data analytics.
Yes, some people even kind of classified these sort
of tricks or techniques to use under information warfare.
So, people were saying that is not,
before you want to actually create these sort of threats or
rumor or disturbance in a particular place, right?
I mean, very often that they create a lot of negative names.
It seems that negatives are from a different direction.
Let's say all the property market is going to be in bubble or whatever.
And people take advantage of the short response that they don't have to,
the market could have a short response and bounce back to normal, right?
But the people creating these noise just need to make money off that short response.
Right.
So the slight drop or slight increase especially a slight drop.
Increase is probably difficult to push the price up but drop is easier to do.
So that's why a lot of these people are purposely using different techniques to kind
of influence the people's behavior
so that they can take advantage of that in the financial market.
Can we use this in other contexts besides FinTech?
Could we influence politics?
Oh yes, that's a very good example.
Because, I mean people are so remote from the candidate, right?
They don't personally know the candidate, right?
Yes.
So, it's through, you know, kind of rumors,
opinions and other people what they are saying
and then by seeing what is the video clip, you know,
on only part of his life or whatever
a snapchat of something and then just kind of exploded.
An example of using social media in an interesting way with data is bombarding
the Russian people with data about how wonderful
Vladimir Putin is reading his own song, giving his own hype.
And you got a great candidate.
People think he is wonderful,
not everybody but a lot of people do.
Yes.
People I know who are from Russia will say,
"no he's really pretty cool.
You should see him wear a shirt on".
Yes.
I don't know if I want to see President Trump without a shirt on but okay,
maybe some people have.
But, at any rate,
so we can get a lot of hype and buzz.
How important is this increasing amount of data,
this better tools for data analytics in FinTech markets?
Now, a lot of those data are actually useful for analysts.
Now, it's really is to look at the pendulum,
do you prefer less data or more data?
A lot people probably say that more data.
More data, yes. At least I have the choice,
I have more information to confer,
to clean it, to filter out all the irrelevant part or the noise part,
at least I have a few instead of have nothing.
There's only one or two source of data then
there's no way for me to know the true picture of it.
So, I think people in general still prefer more data.
And also, once with more data then more people will be able to
built a lot of analytic engines to analyze,
to clean the data, to look at the quality of the data.
But I have heard it said,
at least in the context of regulators,
that regulators were used to thinking we need more data,
we need more data, we need more data.
And that kind of overwhelmed with too much data today.
And so, what do they need when we get so much data,
so much big data,
so many sources of data.
What do we need now?
Yeah. Okay. I think the next thing is data cleaning.
It's really a tedious process.
And very often, a lot of the manager will say,
yes, we do a lot of big data analytics, et cetera.
They overlook the importance of cleaning the data.
And that can cost more than the analytics itself.
It could be, yes.
Use porting, for example,
or the exchange rate.
In some countries, if the government decided to disencourage high frequency trading.
Right.
So, instead of releasing tech data,
they could release second data.
Every second they release,
within the second, they are,
let's say 100 transactions without any time series sequence.
So it's up to you to figure it out,
okay are they purposely randomnize it.
Okay, so, now, it's up to you whether you consider this to be
a fair approach or because for people that doesn't have super computer,
advanced computer because is fair.
That's okay.
That's okay. To me,
whether they could come and take of microseconds or
seconds to do some of the general retail investor than [inaudible].
Frankly, they could give it to me in days, I don't care.
I'm not going to make a decision in less than a day from small traders.
So, yes.
So, that's why regulater do play a role in terms of data quality as well.
So, it depends on how they regulate that source.
And also, for data cleaning or data so called data
"reconstruction" then it's up to the data analytics to look at
that second and try to kind of find out
the sequence of those tech data within the seconds.
A lot of companies talk about big data today,
and a lot of companies talk about analytics.
One of the things that surprises me is how many companies are talking
about big data that actually have small data.
They don't have that much data,
so a client may come to me and say,
I have data on
all the hotel stairs and all my hotels who are members of my frequent flyer club,
and I'm thinking your whole dataset for 10 years is
less than one hour of stock trades on the stock market for one exchange.
So, the amount of data we're talking about that's already pretty clean in
financial markets is huge compared for what it is for a lot of companies.
Right.
So, I would say that makes data analytics
incredibly powerful for financial markets because we have such clean data.
We have such high quality data that it's the collection,
the coalition, and the analysis of that data that it
starts to create a lot of the value in finance markets.
Yes.
Now that's not true a peer-to-peer lending.
It's not true in some other things.
Now, we're looking for social data or insurance, or risk data.
So it depends on which FinTech market we're talking about.
Right, yes. Because, I mean,
especially in the finance market,
we talk about time is critical in terms of the data.
And also, we're talking about the real time data.
Now, in the finance party,
even compared to insurance data and the so-called the stock data,
I mean you don't, for the insurance quote,
you don't change your auto call or your parents life insurance quote every second.
No, I'm not changing my medical prescriptions very often.
So, that's small data by comparison, stockwise.
In the Stockwise, every seconds the trading price on the stock could fluctuate,
no ups and downs within 00.1 percent.
But if you could do arbitraging,
you could do a lot of other techniques then that 00.1
percent could make a lot of difference in the foreign exchange market, as well.
There is something that could make a difference in big data, in health insurance,
or life insurance and that is if we get
wearable technologies that are more mature Internet of things,
I could know what have you ordered,
what have you eaten,
how much have you exercised,
how much insulin did you inject over the past year,
many other things that could affect my analysis of your risk.
You raise a very good point here.
For insurance data on that particular prospect,
we always consider data privacy issue.
It always evoke personal data as compared to the financial data on the stock market.
So, once it involve personal data, basically two approach,
one is the person willing to give you his or her data,
because you are offering him or her some incentive.
Let's say if I offer my half data or my good wear something,
then you give me a little premium.
So, there's the incentive so that I'll be able to, for example,
if I'll let you put in a device in my car to monitor my driving manner,
then probably you would reduce the insurance.
And I believe I'm a good driver then I don't mind you to capture those information.
The other part is I probably don't want to release
my personal information as much as that of exactly how I drive in details but rather,
I'm still releasing some of it for you to do analysis. For example.
Especially if you'll give me a 10 percent discount.
Yes. Exactly.
So, if the price is right,
maybe I'll give up some privacy.
Yes. And also you could use those data to do
further analysis of coming up with new insurance model.
So, you may have a particular insurance model for
particular types of people in a particular occupation, okay, maybe teacher,
I don't know I'm just putting a hypothetical maybe,
after you study, old teachers and professors,
they tend to drive safely or
whatever sort of types of people working in a particular occupation.
I don't know about that. I'm not so sure that that's accurate but, okay.
So, to summarize, we have a lot of data,
particularly, in financial markets.
Data analytics matters.
This is one of the biggest opportunities for data analytics because we have clean data.
We have high quality data for stock trading,
and that makes up big opportunity for data analytics that's interesting.
But we also have a wealth and breadth of data and many other areas of FinTech.
And if we can do better data cleaning and get the quality of that data up,
we may be able to do things that we could never do before.
Right, yes. And also the dissemination of data is very important.
Yes.
Like, every morning you turn on the TV,
there are a lot of so-called financial analysts.
Right.
They're talking about these and all that.
They actually influence the data.
It's a circle, right?
They influence the data they are collecting that data.
So they may influence the sentiment of people.
Proves a lot.
That's why the whole data lifecycle also into the dissemination,
once it disseminates, you collect it again,
clean it again et cetera.
Even if it's fake news or neutral news,
just the reporting of the news can
influence people's sentiment and therefore influence later news.
Right.
Okay. Well, thank you very much.
We look forward to having you in our next session when we'll talk about
AI, DSS, and automation.
Thank you.
Thank you.
Explore our Catalog
Join for free and get personalized recommendations, updates and offers.
Coursera provides universal access to the world’s best education,
partnering with top universities and organizations to offer courses online.
© 2018 Coursera Inc. All rights reserved.
