计算 data.table 每个月不同数量股票的平均月收益

Question

假设我有一个 data.table，priceDT，每天观察多股收益，如下所示：

> priceDT
          Date      Return Share
 1: 2011-01-03  0.04500000   GAI
 2: 2011-01-03 -0.02100000   KDV
 3: 2011-01-04  0.03300000   GAI
 4: 2011-01-04  0.01770000   KDV
 5: 2011-01-05 -0.01742000   GAI
 6: 2011-01-05  0.07900000   KDV
 7: 2011-02-06  0.02400000   GAI
 8: 2011-02-06 -0.02110000   KDV
 9: 2011-02-07 -0.04300000   AFT
10: 2011-02-07  0.01199700   AIP
11: 2011-02-07  0.00551810   ARH
12: 2011-02-07  0.07451101   BIK
13: 2011-02-07 -0.03495597   BLU
14: 2011-02-07 -0.06062462   CGR
15: 2011-02-07 -0.03660000   GAI
16: 2011-02-07 -0.01240000   KDV

我想计算给定月份所有股票的平均月回报率。 所以在 2011 年 1 月，两只股票的平均回报率。 由于份额列，我们知道它只有两股。 第一步是获取当月每股的平均回报。 然后得到当月股票组合的平均回报。 所以一月份，GAI的平均值是0.02019333，KDV的平均值是0.02523333。 因此，该月的平均值为：0.02019333

这就是投资组合回报的逻辑。 我想在 data.table 重复月份的 rest

对于我的样本数据，我想要这样的结果：

portfolio

Date  avg_return
1: 2011-01  0.02271333
2: 2011-02 -0.008700561

数据：

priceDT <- fread(text = "Date, Return, Share
                 2011-01-03,0.045,GAI
                 2011-01-03,-0.021,KDV
                 2011-01-04,0.033,GAI
                 2011-01-04,0.0177,KDV
                 2011-01-05,-0.01742,GAI
                 2011-01-05,0.079,KDV
                 2011-02-06,0.024,GAI
                 2011-02-06,-0.0211,KDV
                 2011-02-07,-0.043,AFT
                 2011-02-07,0.011997,AIP
                 2011-02-07,0.0055181,ARH
                 2011-02-07,0.074511006,BIK
                 2011-02-07,-0.034955973,BLU
                 2011-02-07,-0.060624622,CGR
                 2011-02-07,-0.0366,GAI
                 2011-02-07,-0.0124,KDV
                 ")

portfolio <- fread(text = "Date, avg_return
                   2011-01,0.022713333
                   2011-02,-0.01194431
                   ")

Answer 1

这是另一种方法，尽管我的结果与您的结果不符。

您可以创建一个“年-月”列来对结果进行分组。 按照您的步骤，您可以计算每个月的平均份额（对于每个份额），我们称之为ShareMean 。

然后，您可以计算给定月份所有份额的这些均值的平均值，我们将其称为MonthMean 。

这是你的想法吗？

library(data.table)

priceDT[, YearMonth := list(substr(Date, 1, 7))]
priceDT[, .(ShareMean = mean(Return)), by = c("YearMonth", "Share")][
        , .(MonthMean = mean(ShareMean)), by = "YearMonth"]

Output

   YearMonth    MonthMean
1:   2011-01  0.022713333
2:   2011-02 -0.008700561

Answer 2

你可以直接计算每月的回报，我会这样做：

library(tidyverse)
library(lubridate)

priceDT %>%
mutate(month =  month.abb[month(Date)]) %>%
group_by(month) %>%
summarise(avg_return = mean(Return))

（ month.abb[month(Date)]用于月份缩写，例如 Jan、Feb）

或首先计算给定月份的平均份额：

priceDT %>%
 mutate(month =  month.abb[month(Date)]) %>%
 group_by(month,Share) %>%
 summarise(avg_return = mean(Return))

然后你可以像上面那样计算平均月收益。

Answer 3

priceDT[, mean(Return), by = .(ym = format(Date, "%Y-%m"), Share)
        ][, mean(V1), by = ym]
#         ym           V1
# 1: 2011-01  0.022713333
# 2: 2011-02 -0.008700561

Answer 4

我看不出你是如何迈出这一步来获得所有共享的每月平均回报的……但也许这会让你开始？

#make dates
priceDT[, Date := as.Date( Date ) ]
# step 1: mean by share by month
priceDT[, .(avg_return = mean( Return, na.rm = TRUE) ), 
        by = .( month = format(Date, "%Y-%m"), Share ) ]

但是从这里开始，我看不到提供的portfolio的逻辑......