Problem of missing observations in CAPM estimation

[Chun Sing Chan]

I have following variables from raw data and I want to estimate the return of Stock PG using CAPM model. I found that there are a lot of missing observations due to the date gaps since trading does not occur in the weekend and during public holidays. How can I fix this problem? Should I use a new date variable?

Moreover, after fixing of date, i use the following commands to get the estimation result:

tsset date, d
gen lnPG=ln(PG)
gen rPG=100*(lnPG-L.lnPG)
gen rirf=rPG-rfr
reg rirf r_market

Is it appropriate to use gen rPG=100*(lnPG-L.lnPG) to generate the return of stock PG from its daily close price?
*just ignore hml and smb because I just use CAPM not Fama-Fetch three factor model

where r_market stands for the return of market portfolio, rfr stands for daily return for the Treasury bills (risk-free rate of return), PG is the adjusted close price of stock PG

[Carlo Lazzaro]

Chun;
see -help datetime_business_calendars-.

[Chun Sing Chan]

After looking at the bcal section of the manual, i am still a bit confused which commands should I use to fix this. Can you show the steps that I can do?

[Carlo Lazzaro]

Chun:
see -Datetime business calendars creation- entry and related examples, Stata .pdf manual.

[Chun Sing Chan]

I have used the command below to convert %td date in my dataset into %tb date, but it turns out generating all missing values "(8,232 missing values generated)". Why is it happening?

generate newdate = bofd("simple", date)