Generate cumulative returns over a dynamic period in Power BI

I have a table that contains the open, close, high, low, and adjusted close prices for a stock from 2015 to 2021. My goal is to produce cumulative returns over any date hierarchy member (year/quarter/month/week/daily).
Below is a snapshot of what I want to achieve and and my current results.

Excel attempt
I added 3 additional columns

(1) DailyReturn (using     F3/F2-1    )
(2) Prevalue (using     H2    )
(3) Cumulative Return (using     ((1+J2)*(1+H3))-1    )

Power BI attempt

• Name of table is 'Amazon2'.

• 'Adjusted close' is the column with the values

I created three (3) additional columns -

(1) Index column - starting with 1

(2) PreValue - this is to capture the previous day adjusted close value

CALCULATE(
    MAX(Amazon2[Adj Close]),
    ALL(Amazon2),
    Amazon2[Index] = EARLIER(Amazon2[Index])-1
)

(3) Return - this gives daily report

IF(Amazon2[PreValue] = BLANK(),
    BLANK(),
    (Amazon2[Adj Close]/Amazon2[PreValue])-1
)    

Below is the snapshot of my selections in Power BI:

And my Power BI output:

How can I produce cumulative returns over any date hierarchy level?

Excel working file

Because you are only looking for cumulative return, we can actually short-circuit iterating over each date's return and multiplying. We can just divide the ending adjusted return by the last adjusted return prior to the selected context.

VAR prior =
CALCULATE (
        FIRSTNONBLANKVALUE ( Amazon2[Date], [Adj Close] ),
        OFFSET( -1, ALLSELECTED ( Amazon2[Date] ), ORDERBY( Amazon2[Date], ASC ) )
    )
RETURN
DIVIDE (
    LASTNONBLANKVALUE ( Amazon2[Date], [Adj Close] ) - prior,
    prior
)

This measure takes the last adjusted close in context divided by the last adjusted close prior to the current visual context. If there is no adjusted close data prior to the date range included in the visual, the first adjusted close in the visual will be used as the denominator for all cumulative returns.

I would highly recommend creating a separate calendar/date table to filter the stock returns table by.

E.g.,

Calendar = CALENDARAUTO()

Create a relationship between 'Calendar'[Date] and Amazon2[Date].

Add columns on Calendar for week, month, quarter, and year. Here are some examples, but you can definitely adjust these as needed.

Week = 'Calendar'[Date] - Weekday('Calendar'[Date],3) //First day of weeks (starting on Monday)
Month = FORMAT('Calendar'[Date],"mmm") //Month name formatted to three characters
Quarter = "Qtr"&FORMAT(INT((MONTH('Calendar'[Date])-1)/3)+1,"@") //Quarter formatted as Qtr#
Year = YEAR('Calendar'[Date])