R XTS财务盘中数据 - 计算会话值

Question

我有一个XTS对象，其中包含日内（1分钟OHLC）金融时间内部数据。

如何添加代表当前会话值的列，例如sessionHigh，sessionLow等，到目前为止？

进一步指定：对于任何我想知道到目前为止最高和最低价格的行是什么。在第二天开始时，这应该重置为新一天的第一个数据点的高价和低价。

我如何在R中解决这个问题？如何根据数据的（每日）子集计算新列？

Answer 1

您可以使用纯XTS方法进行此操作：

这种方法是一般的。您可以通过split.xts

中的f参数将拆分更改为几周，几个月，小时

library(lubridate)
# generate some toy data:
set.seed(4)
time <- as.POSIXct(as.character(Sys.Date() + days(0:5)))
time <- rep(time, 5) + sample(x= 1:84000, replace = FALSE, size = 5)
time <- time[order(time)]
x <- xts(runif(length(time)), time)
# Solve your problem:
x.byday <- lapply(split(x, f = "days"), function(x) setNames(merge(x, cummax(x), cummin(x)), c("Close", "Close.runmax", "Close.runmin")))
z <- do.call(rbind, x.byday)
> head(z, 15)
#                          Close Close.runmax Close.runmin
# 2017-11-23 00:12:32 0.26042777    0.2604278   0.26042777
# 2017-11-23 06:28:19 0.72440589    0.7244059   0.26042777
# 2017-11-23 06:51:14 0.90609215    0.9060922   0.26042777
# 2017-11-23 13:40:08 0.94904022    0.9490402   0.26042777
# 2017-11-23 18:58:57 0.07314447    0.9490402   0.07314447
# 2017-11-24 00:12:32 0.75467503    0.7546750   0.75467503
# 2017-11-24 06:28:19 0.28600062    0.7546750   0.28600062
# 2017-11-24 06:51:14 0.10005352    0.7546750   0.10005352
# 2017-11-24 13:40:08 0.95406878    0.9540688   0.10005352
# 2017-11-24 18:58:57 0.41560712    0.9540688   0.10005352
# 2017-11-25 00:12:32 0.45510242    0.4551024   0.45510242
# 2017-11-25 06:28:19 0.97105566    0.9710557   0.45510242
# 2017-11-25 06:51:14 0.58398798    0.9710557   0.45510242
# 2017-11-25 13:40:08 0.96220462    0.9710557   0.45510242
# 2017-11-25 18:58:57 0.76170240    0.9710557   0.45510242

请下次提供一些可再现的数据。

Answer 2

在这里我使用的是每日数据，但是在日内数据中会相同：

library(xts)
data(sample_matrix)
sample_close <- as.xts(sample_matrix[,4,drop=FALSE])
sample_close$session_high <- cummax(sample_close)
head(sample_close)
              Close session_high
2007-01-02 50.11778     50.11778
2007-01-03 50.39767     50.39767
2007-01-04 50.33236     50.39767
2007-01-05 50.33459     50.39767
2007-01-06 50.18112     50.39767
2007-01-07 49.99185     50.39767

Answer 3

虽然XTS具有美好的事物，但我已经朝着整理和tibbles转向。还有新的tibbletime软件包。这就是我要做的。

library(lubridate)
library(tidyverse)
tbl <- tibble(time = ymd_hms(time), value)
tbl %>% 
  mutate(day = date(time)) %>% 
  group_by(day) %>% 
  mutate(max = cummax(value),
         min = cummin(value))
# A tibble: 12 x 5
# Groups:   day [2]
                  time     value        day      max       min
                <dttm>     <dbl>     <date>    <dbl>     <dbl>
 1 2012-03-19 11:31:59 10.554327 2012-03-19 10.55433 10.554327
 2 2012-03-19 12:32:59  9.719728 2012-03-19 10.55433  9.719728
 3 2012-03-19 14:34:59 11.775163 2012-03-19 11.77516  9.719728
 4 2012-03-19 15:36:59 10.187320 2012-03-19 11.77516  9.719728
 5 2012-03-19 16:37:59 11.142526 2012-03-19 11.77516  9.719728
 6 2012-03-19 17:38:59 10.415526 2012-03-19 11.77516  9.719728
 7 2012-03-20 11:36:59 11.229507 2012-03-20 11.22951 11.229507
 8 2012-03-20 12:37:59 10.236680 2012-03-20 11.22951 10.236680
 9 2012-03-20 13:38:59  9.634617 2012-03-20 11.22951  9.634617
10 2012-03-20 14:36:59 11.105144 2012-03-20 11.22951  9.634617
11 2012-03-20 15:37:59  8.906406 2012-03-20 11.22951  8.906406
12 2012-03-20 16:38:59 10.461871 2012-03-20 11.22951  8.906406

数据：

time = c("2012-03-19 11:31:59", "2012-03-19 12:32:59", "2012-03-19 14:34:59", 
         "2012-03-19 15:36:59","2012-03-19 16:37:59", "2012-03-19 17:38:59",
         "2012-03-20 11:36:59","2012-03-20 12:37:59", "2012-03-20 13:38:59",
         "2012-03-20 14:36:59","2012-03-20 15:37:59", "2012-03-20 16:38:59")
set.seed(13)
value <- rnorm(12, mean = 10, sd = 1)

奖金tibbletime：一切都像普通的tibble一样工作，但是tibbletime也很时光。

中的许多不错的功能。

library(tibbletime)
tbl_time <- tbl_time(tbl, index=time)
tbl_time %>% 
  group_by(date(time)) %>%
  mutate(max = cummax(value),
         min = cummin(value))