如何在R中自定义纽约交易时区并转换多时区时间戳？

Alright, let's break down how to solve this problem—focusing first on a base-R solution since that's your priority, then touching on a simpler alternative with the lubridate package.

Core Concept #

The goal is to map the New York trading window (18:00 NY time to 17:00 next NY day) to a virtual "trading timezone" where this interval becomes 00:00 to 23:00. The process boils down to:

• Convert input timestamps (from Japan, Australia, Moscow) to New York time.
• Adjust each timestamp to fit the custom 00:00-23:00 window based on where it falls in the NY trading day.

Base-R Implementation #

This approach uses only built-in R functions, no external packages required. It automatically handles New York's daylight saving time thanks to R's native timezone handling.

Step 1: Define the Conversion Function #

convert_to_custom_trading_tz <- function(timestamps, input_tz) {
  # Convert input timestamps to POSIXct with their original timezone
  ny_time <- as.POSIXct(timestamps, tz = input_tz)
  # Switch timezone to New York
  attr(ny_time, "tzone") <- "America/New_York"
  
  # Extract hour component to determine trading window
  ny_hour <- as.integer(format(ny_time, "%H"))
  
  # Initialize trading date and time variables
  trading_date <- as.Date(ny_time)
  trading_time <- ny_time
  
  # Adjust for timestamps before 18:00 NY time (fall into previous trading day)
  before_18 <- ny_hour < 18
  trading_date[before_18] <- trading_date[before_18] - 1
  trading_time[before_18] <- trading_time[before_18] + 6 * 3600  # Add 6 hours to reach 00:00+
  
  # Adjust for timestamps at/after 18:00 NY time (fall into current trading day)
  after_eq_18 <- !before_18
  trading_time[after_eq_18] <- trading_time[after_eq_18] - 18 * 3600  # Subtract 18 hours to reach 00:00+
  
  # Combine date and time into the custom trading datetime
  custom_trading_dt <- as.POSIXct(
    paste(trading_date, format(trading_time, "%H:%M:%S")),
    tz = "UTC"
  )
  # Label with our custom timezone for clarity (technical storage is UTC)
  attr(custom_trading_dt, "tzone") <- "Custom_Trading_TZ"
  
  return(custom_trading_dt)
}

Step 2: Test with Sample Timestamps #

Let's test with timestamps from each target timezone:

# Test Japan Time (Asia/Tokyo)
japan_times <- as.POSIXct(c("2024-06-01 10:00:00", "2024-06-01 05:00:00"), tz = "Asia/Tokyo")
convert_to_custom_trading_tz(japan_times, "Asia/Tokyo")
# Expected output:
# [1] "2024-05-31 03:00:00 Custom_Trading_TZ" "2024-05-31 23:00:00 Custom_Trading_TZ"

# Test Sydney Time (Australia/Sydney)
sydney_times <- as.POSIXct(c("2024-06-01 12:00:00", "2024-06-01 06:00:00"), tz = "Australia/Sydney")
convert_to_custom_trading_tz(sydney_times, "Australia/Sydney")
# Expected output:
# [1] "2024-05-31 05:00:00 Custom_Trading_TZ" "2024-05-31 21:00:00 Custom_Trading_TZ"

# Test Moscow Time (Europe/Moscow)
moscow_times <- as.POSIXct(c("2024-06-01 20:00:00", "2024-06-01 10:00:00"), tz = "Europe/Moscow")
convert_to_custom_trading_tz(moscow_times, "Europe/Moscow")
# Expected output:
# [1] "2024-06-01 01:00:00 Custom_Trading_TZ" "2024-05-31 17:00:00 Custom_Trading_TZ"

Alternative: Lubridate Package (Simpler Syntax) #

If you're open to using external packages, lubridate simplifies the code with more intuitive date-time functions:

Step 1: Install and Load Lubridate #

install.packages("lubridate")
library(lubridate)

Step 2: Define the Lubridate Conversion Function #

convert_to_custom_trading_tz_lubridate <- function(timestamps, input_tz) {
  # Convert input to NY time
  ny_time <- force_tz(as_datetime(timestamps), tzone = input_tz) %>% 
    with_tz("America/New_York")
  
  # Initialize trading date/time
  trading_date <- date(ny_time)
  trading_time <- ny_time
  
  # Adjust for pre-18:00 NY times
  pre_trading_hours <- hour(ny_time) < 18
  trading_date[pre_trading_hours] <- trading_date[pre_trading_hours] - days(1)
  trading_time[pre_trading_hours] <- trading_time[pre_trading_hours] + hours(6)
  
  # Adjust for post-18:00 NY times
  trading_time[!pre_trading_hours] <- trading_time[!pre_trading_hours] - hours(18)
  
  # Combine and label custom timezone
  custom_trading_dt <- ymd_hms(
    paste(trading_date, format(trading_time, "%H:%M:%S")),
    tz = "UTC"
  )
  attr(custom_trading_dt, "tzone") <- "Custom_Trading_TZ"
  
  return(custom_trading_dt)
}

Key Notes #

• Daylight Saving Time: Both solutions automatically handle NY's DST changes because we convert timestamps to the America/New_York timezone, which R recognizes with built-in DST rules.
• Custom Timezone Label: The Custom_Trading_TZ label is just for readability—under the hood, the datetime is stored in UTC, but the logic aligns perfectly with your desired 00:00-23:00 trading window.
• Base-R vs Lubridate: The base-R version is ideal for environments where external packages aren't allowed, while lubridate offers cleaner, more readable code for everyday use.

内容的提问来源于stack exchange，提问作者Denis