R 创建交易模拟
R create trading simulation
我想写一个函数，代表一个交易模拟。我有一个金融资产价格数据框，我想创建一个基于信号的交易策略。
这是我的数据框：
(open = 开盘价, close = 收盘价, return= 变化率 %)
date open close return signal
<date> <dbl> <dbl> <dbl> <dbl>
1 2015-01-20 213. 211. -1 0
2 2015-01-21 211. 227. 7 1
3 2015-01-22 227. 233. 3 0
4 2015-01-23 234. 233. 0 0
5 2015-01-24 233. 248. 7 1
6 2015-01-25 247. 254. 3 0
7 2015-01-26 254. 273. 8 1
8 2015-01-27 273. 263. -4 0
9 2015-01-28 263. 234. -11 0
10 2015-01-29 233. 234. 0 0
我的交易策略是这样描述的：如果day t上有信号，那么在下面的day t+1上买入到开盘价，卖出到收盘价。在我的示例数据集中，第 2 天有一个信号，所以我在第 3 天买入并以第 3 天的收盘价卖出。我投资 100 美元，所以我在第 3 天获得了 3 美元的收益。下一个信号在第 5 天出现，所以我在第 6 天投资了 103 美元，我得到了额外的 3 美元。这 table 说明了我的交易策略：
date open close return signal trading.strategy capital
<date> <dbl> <dbl> <dbl> <dbl>
1 2015-01-20 213. 211. -1 0 - 100
2 2015-01-21 211. 227. 7 1 - 100
3 2015-01-22 227. 233. 3 0 buy+sell 103
4 2015-01-23 234. 233. 0 0 - 103
5 2015-01-24 233. 248. 7 1 - 103
6 2015-01-25 247. 254. 3 0 buy+sell 106
7 2015-01-26 254. 273. 8 1 - 106
8 2015-01-27 273. 263. -4 0 buy+sell 102
9 2015-01-28 263. 234. -11 0 - 102
10 2015-01-29 233. 234. 0 0 - 102
谁能帮我写一个函数来执行我的交易策略？
这是我的数据：
structure(list(date = structure(c(16455, 16456, 16457, 16458,
16459, 16460, 16461, 16462, 16463, 16464), class = "Date"), open = c(212.91,
211.38, 227.32, 233.52, 232.7, 247.35, 254.08, 273.17, 263.35,
233.35), close = c(211.32, 226.9, 233.41, 232.88, 247.85, 253.72,
273.47, 263.48, 233.91, 233.51), return = c(-1, 7, 3, 0, 7, 3,
8, -4, -11, 0), signal = c(0, 1, 0, 0, 1, 0, 1, 0, 0, 0)), class = c("tbl_df",
"tbl", "data.frame"), row.names = c(NA, -10L))
你可以这样做：
library(tidyverse)
df %>%
mutate(trading.strategy = if_else(lag(signal) == 1, "buy+sell", "-")) %>%
filter(trading.strategy == "buy+sell") %>%
mutate(capital = 100 + cumsum(return)) %>%
right_join(df) %>%
fill(capital) %>%
mutate(capital = if_else(is.na(capital), 100, capital),
trading.strategy = if_else(is.na(trading.strategy), "-", trading.strategy))
date open close return signal trading.strategy capital
<date> <dbl> <dbl> <dbl> <dbl> <chr> <dbl>
1 2015-01-20 213. 211. -1 0 - 100
2 2015-01-21 211. 227. 7 1 - 100
3 2015-01-22 227. 233. 3 0 buy+sell 103
4 2015-01-23 234. 233. 0 0 - 103
5 2015-01-24 233. 248. 7 1 - 103
6 2015-01-25 247. 254. 3 0 buy+sell 106
7 2015-01-26 254. 273. 8 1 - 106
8 2015-01-27 273. 263. -4 0 buy+sell 102
9 2015-01-28 263. 234. -11 0 - 102
10 2015-01-29 233. 234. 0 0 - 102
library(dplyr)
initial_capital <- 10000
df %>%
mutate(
trade = ifelse(lag(signal, default = 0), 1, 0),
trading.strategy = ifelse(trade, "buy+sell", "-"),
days_return = trade * (close - open) / (open),
cum_return = cumsum(days_return),
capital = initial_capital * (1 + cum_return)
) %>%
select(-trade,-return)
# A tibble: 10 x 8
date open close signal trading.strategy days_return cum_return capital
<date> <dbl> <dbl> <dbl> <chr> <dbl> <dbl> <dbl>
1 2015-01-20 213. 211. 0 - 0 0 10000
2 2015-01-21 211. 227. 1 - 0 0 10000
3 2015-01-22 227. 233. 0 buy+sell 0.0268 0.0268 10268.
4 2015-01-23 234. 233. 0 - 0 0.0268 10268.
5 2015-01-24 233. 248. 1 - 0 0.0268 10268.
6 2015-01-25 247. 254. 0 buy+sell 0.0258 0.0525 10525.
7 2015-01-26 254. 273. 1 - 0 0.0525 10525.
8 2015-01-27 273. 263. 0 buy+sell -0.0355 0.0171 10171.
9 2015-01-28 263. 234. 0 - 0 0.0171 10171.
10 2015-01-29 233. 234. 0 - 0 0.0171 10171.
只是在开发@Lennyy 策略，请不要将其视为一个独立的解决方案：
library(tidyverse)
CPTL = 100
right_join(
filter(dat, lag(signal == 1)) %>%
mutate(
trading.strategy = 'buy+sell',
capital = CPTL + cumsum(return / 100) * CPTL
),
dat
) %>%
mutate(
capital = replace_na(fill(., capital)$capital, CPTL),
trading.strategy = replace_na(trading.strategy, '-')
)
# A tibble: 10 x 7
# date open close return signal trading.strategy capital
# <date> <dbl> <dbl> <dbl> <dbl> <chr> <dbl>
# 1 2015-01-20 213. 211. -1 0 - 100
# 2 2015-01-21 211. 227. 7 1 - 100
# 3 2015-01-22 227. 233. 3 0 buy+sell 103
# 4 2015-01-23 234. 233. 0 0 - 103
# 5 2015-01-24 233. 248. 7 1 - 103
# 6 2015-01-25 247. 254. 3 0 buy+sell 106
# 7 2015-01-26 254. 273. 8 1 - 106
# 8 2015-01-27 273. 263. -4 0 buy+sell 102
# 9 2015-01-28 263. 234. -11 0 - 102
#10 2015-01-29 233. 234. 0 0 - 102
我想写一个函数，代表一个交易模拟。我有一个金融资产价格数据框，我想创建一个基于信号的交易策略。
这是我的数据框：
(open = 开盘价, close = 收盘价, return= 变化率 %)
date open close return signal
<date> <dbl> <dbl> <dbl> <dbl>
1 2015-01-20 213. 211. -1 0
2 2015-01-21 211. 227. 7 1
3 2015-01-22 227. 233. 3 0
4 2015-01-23 234. 233. 0 0
5 2015-01-24 233. 248. 7 1
6 2015-01-25 247. 254. 3 0
7 2015-01-26 254. 273. 8 1
8 2015-01-27 273. 263. -4 0
9 2015-01-28 263. 234. -11 0
10 2015-01-29 233. 234. 0 0
我的交易策略是这样描述的：如果day t上有信号，那么在下面的day t+1上买入到开盘价，卖出到收盘价。在我的示例数据集中，第 2 天有一个信号，所以我在第 3 天买入并以第 3 天的收盘价卖出。我投资 100 美元，所以我在第 3 天获得了 3 美元的收益。下一个信号在第 5 天出现，所以我在第 6 天投资了 103 美元，我得到了额外的 3 美元。这 table 说明了我的交易策略：
date open close return signal trading.strategy capital
<date> <dbl> <dbl> <dbl> <dbl>
1 2015-01-20 213. 211. -1 0 - 100
2 2015-01-21 211. 227. 7 1 - 100
3 2015-01-22 227. 233. 3 0 buy+sell 103
4 2015-01-23 234. 233. 0 0 - 103
5 2015-01-24 233. 248. 7 1 - 103
6 2015-01-25 247. 254. 3 0 buy+sell 106
7 2015-01-26 254. 273. 8 1 - 106
8 2015-01-27 273. 263. -4 0 buy+sell 102
9 2015-01-28 263. 234. -11 0 - 102
10 2015-01-29 233. 234. 0 0 - 102
谁能帮我写一个函数来执行我的交易策略？
这是我的数据：
structure(list(date = structure(c(16455, 16456, 16457, 16458,
16459, 16460, 16461, 16462, 16463, 16464), class = "Date"), open = c(212.91,
211.38, 227.32, 233.52, 232.7, 247.35, 254.08, 273.17, 263.35,
233.35), close = c(211.32, 226.9, 233.41, 232.88, 247.85, 253.72,
273.47, 263.48, 233.91, 233.51), return = c(-1, 7, 3, 0, 7, 3,
8, -4, -11, 0), signal = c(0, 1, 0, 0, 1, 0, 1, 0, 0, 0)), class = c("tbl_df",
"tbl", "data.frame"), row.names = c(NA, -10L))
你可以这样做：
library(tidyverse)
df %>%
mutate(trading.strategy = if_else(lag(signal) == 1, "buy+sell", "-")) %>%
filter(trading.strategy == "buy+sell") %>%
mutate(capital = 100 + cumsum(return)) %>%
right_join(df) %>%
fill(capital) %>%
mutate(capital = if_else(is.na(capital), 100, capital),
trading.strategy = if_else(is.na(trading.strategy), "-", trading.strategy))
date open close return signal trading.strategy capital
<date> <dbl> <dbl> <dbl> <dbl> <chr> <dbl>
1 2015-01-20 213. 211. -1 0 - 100
2 2015-01-21 211. 227. 7 1 - 100
3 2015-01-22 227. 233. 3 0 buy+sell 103
4 2015-01-23 234. 233. 0 0 - 103
5 2015-01-24 233. 248. 7 1 - 103
6 2015-01-25 247. 254. 3 0 buy+sell 106
7 2015-01-26 254. 273. 8 1 - 106
8 2015-01-27 273. 263. -4 0 buy+sell 102
9 2015-01-28 263. 234. -11 0 - 102
10 2015-01-29 233. 234. 0 0 - 102
library(dplyr)
initial_capital <- 10000
df %>%
mutate(
trade = ifelse(lag(signal, default = 0), 1, 0),
trading.strategy = ifelse(trade, "buy+sell", "-"),
days_return = trade * (close - open) / (open),
cum_return = cumsum(days_return),
capital = initial_capital * (1 + cum_return)
) %>%
select(-trade,-return)
# A tibble: 10 x 8
date open close signal trading.strategy days_return cum_return capital
<date> <dbl> <dbl> <dbl> <chr> <dbl> <dbl> <dbl>
1 2015-01-20 213. 211. 0 - 0 0 10000
2 2015-01-21 211. 227. 1 - 0 0 10000
3 2015-01-22 227. 233. 0 buy+sell 0.0268 0.0268 10268.
4 2015-01-23 234. 233. 0 - 0 0.0268 10268.
5 2015-01-24 233. 248. 1 - 0 0.0268 10268.
6 2015-01-25 247. 254. 0 buy+sell 0.0258 0.0525 10525.
7 2015-01-26 254. 273. 1 - 0 0.0525 10525.
8 2015-01-27 273. 263. 0 buy+sell -0.0355 0.0171 10171.
9 2015-01-28 263. 234. 0 - 0 0.0171 10171.
10 2015-01-29 233. 234. 0 - 0 0.0171 10171.
只是在开发@Lennyy 策略，请不要将其视为一个独立的解决方案：
library(tidyverse)
CPTL = 100
right_join(
filter(dat, lag(signal == 1)) %>%
mutate(
trading.strategy = 'buy+sell',
capital = CPTL + cumsum(return / 100) * CPTL
),
dat
) %>%
mutate(
capital = replace_na(fill(., capital)$capital, CPTL),
trading.strategy = replace_na(trading.strategy, '-')
)
# A tibble: 10 x 7
# date open close return signal trading.strategy capital
# <date> <dbl> <dbl> <dbl> <dbl> <chr> <dbl>
# 1 2015-01-20 213. 211. -1 0 - 100
# 2 2015-01-21 211. 227. 7 1 - 100
# 3 2015-01-22 227. 233. 3 0 buy+sell 103
# 4 2015-01-23 234. 233. 0 0 - 103
# 5 2015-01-24 233. 248. 7 1 - 103
# 6 2015-01-25 247. 254. 3 0 buy+sell 106
# 7 2015-01-26 254. 273. 8 1 - 106
# 8 2015-01-27 273. 263. -4 0 buy+sell 102
# 9 2015-01-28 263. 234. -11 0 - 102
#10 2015-01-29 233. 234. 0 0 - 102