Question

我有一个数据框，我想通过删除一些抵消线（盒装位置）和做一些净额来清理。

这是源表：

    Type  Name     Strike  Maturity    Nominal
    Call  Amazon    10     10/12/2018  1000
    Put   Amazon    10     10/12/2018  1000
    Call  Ebay      8      2/8/2018    800
    Put   Ebay      8      2/8/2018    500
    Call  Facebook  5      5/5/2018    900
    Call  Google    2      23/4/2018   250
    Put   Google    2      23/4/2018   350
    Call  Microsoft 2      19/3/2018   250
    Put   Microsoft 2.5    19/3/2018   350
    Put   Ebay      8      2/8/2018    100

代码的结果在这里：

    Type  Name      Strike  Maturity   Nominal
    Call  Ebay      8       2/8/2018   200
    Call  Facebook  5       5/5/2018   900
    Put   Google    2       23/4/2018  100
    Call  Microsoft 2       19/3/2018  250
    Put   Microsoft 2.5     19/3/2018  350

我试图在R中编写一个执行这3项任务的代码：

1 //删除所有相互抵消的对。相互抵消的一对是符合这两个标准的一对：

具有相同名称，打击，成熟度和名义的2行。
1行是＆＃34; Call＆＃34;而另一个是＆＃34; Put＆＃34;

示例：2＆＃34;亚马逊＆＃34;从表中删除的行

2 //对不完全相互抵消的线路进行标称净值。 一对不完全相互抵消的对是符合以下两个标准的一对：

具有相同名称，打击和成熟度但名义不同的2行
1行是＆＃34; Call＆＃34;而另一个是＆＃34; Put＆＃34;

示例：2＆＃34; Ebay＆＃34;线路已在通话或2＆＃34; Google＆＃34;在Put上被净获的线。

3 //不要在所有其他行上做任何事情

示例：2＆＃34; Microsoft＆＃34;线。他们有不同的罢工，所以根本不应该进行净额结算

请参阅下面我的第一次尝试。我的想法是先创建一个带有唯一键的新列，然后按字母顺序排序，然后逐个测试每一行。我发现它非常费力，所以我想知道是否有人可以帮我找到更简单有效的解决方案？非常感谢！

library(data.table)

dt <- data.table(Type=c("Call", "Put", "Call", "Put", "Call", "Call", "Put", "Call", "Put","Put"),
                 Name=c("Amazon", "Amazon", "Ebay", "Ebay", "Facebook", "Google", "Google", "Microsoft", "Microsoft","Ebay"),
                 Strike=c(10,10,8,8,5,2,2,2,2.5,8),
                 Maturity=c("10/12/2018", "10/12/2018", "2/8/2018", "2/8/2018", "5/5/2018", "23/4/2018", "23/4/2018", "19/3/2018", "19/3/2018","2/8/2018),
                 Nominal=c(1000,1000,800,500,900,250,350,250,35,100))

##idea
dt$key <- paste(dt$Name,dt$Strike,dt$Maturity)
dt[order(dt$key,decreasing = FALSE),]
dt$Type2 <- ifelse(dt$Type = "Call",1,0)

#for each line k, test value in the column "Key" and the column "Type2":
#if key(k) = key(k+1) and Type2(k)+Type2(k+1)=1 then 
    #if Nominal (k)> Nominal (k+1), delete the line k+1 and do the netting on nominal of the line k
    #else Nomnial (k+1)< Nominal (k), delete the line k and do the netting on nominal of the line k+1
#next k

dt <- dt[dt$Nominal!=0,]
dt$key <- NULL

在推荐的想法之后，我尝试了dcast解决方案，但看起来它没有做正确的网络，如下所示：

> dt <- data.table(Type=c("Call", "Put", "Call", "Put", "Call", "Call", "Put", "Call", "Put","Put"),
+                  Name=c("Amazon", "Amazon", "Ebay", "Ebay", "Facebook", "Google", "Google", "Microsoft", "Microsoft","Ebay"),
+                  Strike=c(10,10,8,8,5,2,2,2,2.5,8),
+                  Maturity=c("10/12/2018", "10/12/2018", "2/8/2018", "2/8/2018", "5/5/2018", "23/4/2018", "23/4/2018", "19/3/2018", "19/3/2018","2/8/2018"),
+                  Nominal=c(1000,1000,800,500,900,250,350,250,350,100))
> dcast(dt, Name + Maturity + Strike ~ Type, value.var="Nominal", fill = 0)[, Net := Call - Put][Net != 0]
Aggregate function missing, defaulting to 'length'
        Name  Maturity Strike Call Put Net
1:      Ebay  2/8/2018    8.0    1   2  -1
2:  Facebook  5/5/2018    5.0    1   0   1
3: Microsoft 19/3/2018    2.0    1   0   1
4: Microsoft 19/3/2018    2.5    0   1  -1

Answer 1

这是一个tidyverse解决方案。基本上，由于您要对具有相同Name，Strike和Maturity的所有行进行分组，因此我认为将Call和Put转换为实际最简单数字并使用summarise。您的特殊偏移情况实际上只是删除了总数最终为0的网络案例。

方法是：

使用Put和Nominal将ifelse转换为mutate的负值，
使用group_by和summarise将群组缩减为每个群组的单个值，
使用filter，
替换Type列并将负值设为正值。

代码：

library(tidyverse)
tbl <- read_table2(
  "Type  Name     Strike  Maturity    Nominal
  Call  Amazon    10     10/12/2018  1000
  Put   Amazon    10     10/12/2018  1000
  Call  Ebay      8      2/8/2018    800
  Put   Ebay      8      2/8/2018    500
  Call  Facebook  5      5/5/2018    900
  Call  Google    2      23/4/2018   250
  Put   Google    2      23/4/2018   350
  Call  Microsoft 2      19/3/2018   250
  Put   Microsoft 2.5    19/3/2018   350
  Put   Ebay      8      2/8/2018    100"
)

tbl %>%
  mutate(actual = ifelse(Type == "Call", Nominal, -Nominal)) %>%
  group_by(Name, Strike, Maturity) %>%
  summarise(Net = sum(actual)) %>%
  filter(Net != 0) %>%
  mutate(
    Type = ifelse(Net > 0, "Call", "Put"),
    Net = abs(Net)
    )
# A tibble: 5 x 5
# Groups:   Name, Strike [5]
  Name      Strike Maturity    Net Type 
  <chr>      <dbl> <chr>     <int> <chr>
1 Ebay        8.00 2/8/2018    200 Call 
2 Facebook    5.00 5/5/2018    900 Call 
3 Google      2.00 23/4/2018   100 Put  
4 Microsoft   2.00 19/3/2018   250 Call 
5 Microsoft   2.50 19/3/2018   350 Put

在数据帧中净额结算

1 个答案: