🔍
Course Introduction | Portfolio Management using Machine Learning : Hierarchical Risk Parity - YouTube
Channel: Quantra
[0]
Welcome to this course on portfolio management
[2]
using machine learning.
[5]
After completing this video, you will:
[7]
Have an overview of what you will achieve
[9]
by the end of this course.
[11]
Get a brief idea of how exactly machine
[14]
learning can help in portfolio management.
[17]
And know what you will be learning
[18]
in the course.
[21]
After completing this course, you will
[23]
have a hands-on experience of how
[25]
machine learning is used in
[26]
portfolio management.
[29]
You will be able to use it to calculate
[31]
the optimal allocation of capital to assets
[34]
in your own selected portfolio.
[37]
There are many ways to use
[38]
machine learning techniques
[40]
for portfolio management,
[41]
but we will focus on
[42]
hierarchical clustering in this course.
[45]
Hierarchical clustering can be implemented
[47]
universally, thus, you can use it on
[50]
any asset class or geography.
[53]
Why is portfolio management important?
[57]
Let’s say that your portfolio consisted
[59]
of Amazon and Apple in the year 2000.
[63]
These are both tech-related stocks.
[66]
When the tech bubble burst, your portfolio
[68]
value would have declined sharply.
[71]
Amazon declined from $90 to $20 in a year.
[76]
And Apple declined from $1 to $0.2.
[80]
Your portfolio would have gone down by 78%.
[85]
One of the main ideas of portfolio management
[87]
can be summed up by a beautiful quote.
[90]
“Do not put all your eggs in one basket.”
[95]
So you decide on picking one stock each
[98]
from technology, FMCG, finance and the
[101]
automotive sector.
[104]
This is also called creating a
[105]
diversified portfolio.
[108]
You should also look at the role of risk
[110]
in your portfolio.
[112]
Tesla is a highly volatile stock.
[116]
A rise or fall of 10% in Tesla’s prices
[118]
is not considered extreme.
[121]
But stocks like Verizon or AT&T hardly
[124]
move beyond 5%.
[127]
Thus, if you have two high volatile
[129]
and two low volatile stocks,
[132]
you will see that your portfolio is
[134]
driven by high volatility stocks.
[137]
There are various ways to overcome
[139]
this limitation.
[141]
You will go through portfolio management
[143]
techniques like inverse volatility
[146]
and critical line algorithm.
[148]
But these techniques have some
[150]
inherent limitations.
[152]
There can be concentration of risk
[154]
and high computing requirements.
[157]
So, we turn towards machine learning
[159]
for a solution.
[162]
How can machine learning help in
[163]
portfolio management?
[167]
Machines can find patterns that humans can’t.
[170]
Machines are beating humans at
[172]
strategy games like Chess and Go.
[176]
Then, why not use advanced techniques
[178]
for portfolio management?
[181]
Unsupervised learning, a way of machine
[183]
learning, tries to group similar assets
[186]
to find trading opportunities using
[188]
the assets’ features.
[190]
Assume that you pass the daily return
[192]
as a feature of 4 stocks, Google, Nvidia,
[196]
Electronic Arts, and Amazon.
[199]
The hierarchical clustering algorithm
[201]
puts Google and Amazon in one group,
[203]
while Nvidia and Electronic Arts are put
[205]
in another.
[207]
This grouping is done based on similarities
[210]
between the individual stocks.
[213]
The same information later becomes crucial
[215]
while deciding the weight allocation
[217]
for these stocks.
[219]
This seems basic enough and can be done
[220]
manually.
[222]
But we will pass many more features
[224]
and our asset universe will be large.
[227]
How is hierarchical clustering useful
[230]
in portfolio management?
[232]
Many studies have concluded that
[234]
hierarchical clustering produces better
[236]
results as compared to the Critical
[238]
line algorithm or the inverse
[240]
volatility approach.
[243]
This is because it does not concentrate
[245]
on a few assets.
[247]
It also provides a portfolio with
[249]
the least variation in its returns.
[252]
The course also has a capstone project
[255]
so that you can assimilate and apply
[257]
your learnings from this course.
[259]
To enable you to implement the
[261]
portfolio management technique, we will
[263]
also provide a live trading template.
[266]
You can use this template to apply
[268]
hierarchical clustering on your
[270]
selected assets and use it to paper trade
[273]
with real-time market data.
[275]
If you are satisfied with the results,
[277]
you can move further and live trade as well.
[281]
All the strategy codes and data files
[283]
will be provided to you in the
[284]
last section of the course.
[287]
You can download them on your local system too.
[291]
You can use them to tweak and modify
[292]
the code to truly make it your own trading
[295]
strategy.
[297]
All it takes is a few hours and
[298]
you will be on your way to apply a
[300]
robust portfolio management technique.
[304]
Let’s get started then!
Most Recent Videos:
You can go back to the homepage right here: Homepage