Py 79 Diversifiable and Non Diversifiable Risk - YouTube

Ok, people!

With this lesson, we’ll complete the section in which we studied the calculation of financial

risk.

Our task here is to disaggregate a portfolio’s diversifiable and non-diversifiable risk.

The portfolio we’ll use is a simple one containing only two securities, but the exact

same mechanics can be applied when working with a larger portfolio.

To start, we will need the weights of the two assets.

Given this is an equally weighted portfolio of the two stocks, we will need a NumPy array

containing twice the value of 0.5.

We already created one and called it “weights”.

The values of this array can be accessed after we index the zero and the first position in

brackets.

Ok, the portfolio weights are ready.

One way to estimate the annual diversifiable risk is to obtain the portfolio variance,

and then subtract the weighted annual variance of each stock.

This occasion will allow us to emphasize the difference between creating an “ND” array

with single or double brackets.

We explained in one of our earlier videos that practically each pair of brackets around

the name of the column will add one dimension to the NumPy array.

For the sake of argument, let’s use this structure to create the annual variance of

each company.

What record the values of the variances of P&G and Beiersdorf not as floats, but as single

elements in 1 by 1 matrices – that is, two-dimensional arrays.

For this reason, we should expect a wrong output when we calculate the unsystematic

or also known as diversifiable risk.

The formula we have typed is correct.

Well, we don’t see an error, but the answer is strange anyway, right?

Instead of a float, we obtain a vector with no numbers.

This little exercise was to remind you to be careful when creating NumPy objects.

Let’s see two techniques that could help you overcome this issue.

Since the variance of Procter & Gamble is stored in a 1 by 1 matrix, it will be a single

value.

If you apply the well-known float function, you will turn this value into a float.

Nice!

Alternatively, when you assign a value to “PG, var, A”, you can use single brackets

when you indicate you will use data from the P.G. column.

This will automatically store the output as a float value.

The “data type float 64” line will disappear, and you will see more digits after the decimal

point.

Great!

To be consistent, let’s apply the second method to the annual variance of Beiersdorf

as well.

“BEI, var, A” will be a floating point.

Good!

Can we try to re-estimate the diversifiable, or also known as unsystematic, risk?

Yes, we can.

And we obtain a float whose transformation shows it is something smaller than a percent.

Great!

The remaining part of the portfolio variance is the systematic risk.

As you can see, regardless of whether we subtract the unsystematic risk from the whole variance,

or we sum the weighted annual variances, we obtain the same value.

The check we make in the last cell further confirms our results – we obtain the boolean

value “True” in the output as we verify the equality between the two variables, “SR

1”, and “SR 2”.

Ok.

Great!

We are doing well!

In the next chapter, we will start a new topic – regression analysis.

See you there!

And… thanks for watching!