Calculate daily log return in R-CodePudding

I am trying to calculate the daily log return based on daily closing prices for different stocks. In total, I have a dataset of 4,000 stocks covering a time period of one year.

Here is some trial data to illustrate my current data basis:

df <- tibble(
  Date = c(
    '01.11.2019',
    '04.11.2019',
    '05.11.2019',
    '06.11.2019',
    '07.11.2019',
    '08.11.2019'
  ),
  ACCR.PK = c(0.0035, 0.003, 0.0035, 0.0057, 0.0032, 0.0032),
  SWGI.PK = c(0.51, 0.51, 0.51, 0.51, 0.51, 0.51),
  HURC.OQ = c(35.53, 35.62, 35.76, 35.52, 35.6, 36)
) %>% 
  mutate(Date = Date %>% 
           as.Date(format = "%d.%m.%Y"))

df <- pivot_longer(df, -Date, names_to = "ticker", values_to = "price") 
df <- group_by(ticker)

Which should look like this:

I am trying to calculate the daily log returns for each stock. I tried this function:

lrtn=diff(log(df$price))
Log.return.df <- data.frame(lrtn)

This function does indeed calculate the log returns, but it seems to delete the first row (since it is not possible to delete the first log without the prior data). I still want to keep the first row though (e.g. indicated as zero or N/A).

Also, the formula seems to not differentiate between the stocks.

Can somebody help me?

CodePudding user response：

You can do,

library(dplyr)

df %>% 
 group_by(ticker) %>% 
 mutate(lrtn = c(NA, diff(log(price)))) %>% 
 select(lrtn)

Adding missing grouping variables: `ticker`
# A tibble: 18 × 2
# Groups:   ticker [3]
   ticker      lrtn
   <chr>      <dbl>
 1 ACCR.PK NA      
 2 SWGI.PK NA      
 3 HURC.OQ NA      
 4 ACCR.PK -0.154  
 5 SWGI.PK  0      
 6 HURC.OQ  0.00253
 7 ACCR.PK  0.154  
 8 SWGI.PK  0      
 9 HURC.OQ  0.00392
10 ACCR.PK  0.488  
11 SWGI.PK  0      
12 HURC.OQ -0.00673
13 ACCR.PK -0.577  
14 SWGI.PK  0      
15 HURC.OQ  0.00225
16 ACCR.PK  0      
17 SWGI.PK  0      
18 HURC.OQ  0.0112