How to calculate pro rata portfolio weights based on values within groups?

Question

I want to calculate portfolio weights based on their Fund value relative to the total fund value in each portfolio.

I have a 50,000 x 4 data frame. The fund value column should determine the weight of each fund within each draw.

 draw Fund.ID Fund.Value..mn.USD. Net.Multiple..X.
   (int)   (int)               (dbl)            (dbl)
1      1   10678              1963.8             1.29
2      1    8812              9400.0             1.61
3      1    7236              7525.0             1.58
4      1   12702              5979.7             1.40
5      1   13715              3510.0             1.70
6      2    2060               334.8             1.42
7      2    2059               250.0             1.47
8      2    1151               202.0             2.12
9      2    1812               500.0             0.11
10     2    6822               151.2             2.09

For draw 1, the fund in row 2 should have the highest weight (33.12%), the fund in row 3 should have the second highest weight (26.52%) and the fund in row 1 the lowest weight (6.92%). The sum of weights from row 1-5 should equal 100%.

How can I create such a weight vector for 1,000 draws? I then want to multiply each weight with the respective Net Multiple (weight(fund1)*Net Multiple(fund1)) .

Answer 1

We can use dplyr to get the fund weight, also, let's call your data.frame df , and make the column names single words so they're a bit easier to deal with.

library(dplyr)
head(df)
  Draw  Fund  Value Multiple
1    1 10678 1963.8     1.29
2    1  8812 9400.0     1.61
3    1  7236 7525.0     1.58
4    1 12702 5979.7     1.40
5    1 13715 3510.0     1.70
6    2  2060  334.8     1.42

df <- df %>% group_by(Draw) %>% mutate(FundWeight = Value / sum(Value))
head(df)

Source: local data frame [6 x 5]
Groups: Draw [2]

   Draw  Fund  Value Multiple FundWeight
  (int) (int)  (dbl)    (dbl)      (dbl)
1     1 10678 1963.8     1.29 0.06920027
2     1  8812 9400.0     1.61 0.33123668
3     1  7236 7525.0     1.58 0.26516553
4     1 12702 5979.7     1.40 0.21071234
5     1 13715 3510.0     1.70 0.12368518
6     2  2060  334.8     1.42 0.23282337

#multiply columns
df$WeightMultiple <- df$FundWeight * df$Multiple
df

Source: local data frame [10 x 6]
Groups: Draw [2]

    Draw  Fund  Value Multiple FundWeight WeightMultiple
   (int) (int)  (dbl)    (dbl)      (dbl)          (dbl)
1      1 10678 1963.8     1.29 0.06920027     0.08926835
2      1  8812 9400.0     1.61 0.33123668     0.53329105
3      1  7236 7525.0     1.58 0.26516553     0.41896154
4      1 12702 5979.7     1.40 0.21071234     0.29499727
5      1 13715 3510.0     1.70 0.12368518     0.21026481
6      2  2060  334.8     1.42 0.23282337     0.33060918
7      2  2059  250.0     1.47 0.17385257     0.25556328
8      2  1151  202.0     2.12 0.14047288     0.29780250
9      2  1812  500.0     0.11 0.34770515     0.03824757
10     2  6822  151.2     2.09 0.10514604     0.21975522

Answer 2

You can use data.table ,

library(data.table) 
setDT(df)[, weight := prop.table(Fund.Value), by = draw]
df
#    draw Fund.ID Fund.Value Net.Multiple..X.     weight
# 1:    1   10678     1963.8             1.29 0.06920027
# 2:    1    8812     9400.0             1.61 0.33123668
# 3:    1    7236     7525.0             1.58 0.26516553
# 4:    1   12702     5979.7             1.40 0.21071234
# 5:    1   13715     3510.0             1.70 0.12368518
# 6:    2    2060      334.8             1.42 0.23282337
# 7:    2    2059      250.0             1.47 0.17385257
# 8:    2    1151      202.0             2.12 0.14047288
# 9:    2    1812      500.0             0.11 0.34770515
#10:    2    6822      151.2             2.09 0.10514604