累积和之字形指标

Question

以下代码提供了示例数据：

library(TTR)
set.seed(15)
r <- rnorm(1000, 0, .01)
P_1 <- 100
P <- P_1*cumprod(1+r)
zz <- ZigZag(P, change = 5, percent = TRUE)
set.seed(15)
volume <- round(runif(1000, 50, 550), digits = 0)
data <- as.data.frame(cbind(P, zz, volume))
plot(P, type = "l")
lines(zz, col = "red")

最后我想在新列中创建累积的音量总和，当曲折线（zz）改变方向时会发生重置。我曾尝试使用s <- sign(diff(data$zz, lag = 1))，这会显示那些转折点，但却无法使用cumsum。

Answer 1

以下是使用dplyr的解决方案：

library(dplyr)

data %>%
  mutate(
    zz_up = (zz - lag(zz) > 0),
    zz_switch = zz_up != lag(zz_up),
    zz_switch = ifelse(is.na(zz_switch), FALSE, zz_switch),
    group = cumsum(zz_switch)
    ) %>%
  group_by(group) %>%
  mutate(cum_volume = cumsum(volume)) 

Answer 2

尝试RcppRoll：

代码

Vectorize(require)(package = c("magrittr", "dplyr", "RcppRoll"),
                               char = TRUE)

data %<>%
  # Create difference for ZigZag
  mutate(diffZZ = c(0,diff(zz))) %>% 
  # Use it as a group
  group_by(diffZZ) %>% 
  # Use RcppRoll to compute that sum
  mutate(sumVolByDiff = roll_sum(x = volume, n = 2, fill = NA)) %>% 
  # Clean / not important
  ungroup()

预览

> head(data)
Source: local data frame [6 x 5]

         P       zz volume   diffZZ sumVolByDiff
     (dbl)    (dbl)  (dbl)    (dbl)        (dbl)
1 100.2588 100.2588    351 0.000000           NA
2 102.0947 100.5596    148 0.300785          523
3 101.7480 100.8604    533 0.300785         1077
4 102.6608 101.1612    375 0.300785          609
5 103.1618 101.4620    234 0.300785          692
6 101.8668 101.7627    544 0.300785          938