R

KDJ指标择时交易策略分析

量化投资
knitr::opts_chunk$set(echo = TRUE, 
                      warning = FALSE, 
                      message = FALSE, 
                      fig.pos = 'H', 
                      fig.align = 'center', 
                      fig.width = 6.5, 
                      fig.height = 4, 
                      out.width = "90%", 
                      width = "90%")

引言

技术分析是金融市场中常用的分析方法，其中KDJ指标是一种重要的随机指标，能够反映价格波动的强弱、超买超卖现象以及市场趋势变化。本研究旨在通过R语言实现基于KDJ指标的股票择时交易策略，并通过历史数据回测寻找最佳参数组合。

研究方法

数据获取与处理

我们将使用quantmod包获取股票数据，并使用quantstrat包进行策略回测。首先加载所需的包：

# 加载必要的包
library(quantmod)
library(quantstrat)
library(eTTR)
library(PerformanceAnalytics)
library(ggplot2)
library(dplyr)
library(tibble)
library(scales)
library(gridExtra)
# 加载环境重置函数
source("/Users/matrixspk/My-Sites/r-finance/assets/code/reset_strategy_env.R")
# 加载计算胜率的函数
source("/Users/matrixspk/My-Sites/r-finance/assets/code/calculate_portfolio_win_rates.R")
source("/Users/matrixspk/My-Sites/r-finance/assets/code/generateSimpleSignalChain.R")

接下来，我们获取苹果公司股票的历史数据作为研究对象：

# 设置获取数据的起始和结束日期
initDate <- as.Date("2017-12-31")
startdate.st <- as.Date("2018-01-01")
enddate.st <- as.Date("2023-06-01")
# 获取苹果公司股票数据
getSymbols("AAPL", src = "yahoo", from = startdate.st, to = enddate.st)

## [1] "AAPL"

colnames(AAPL) <- c("Open", "High", "Low", "Close", "Volume", "Adjusted")
# 查看数据结构
head(AAPL)

##               Open    High     Low   Close    Volume Adjusted
## 2018-01-02 42.5400 43.0750 42.3150 43.0650 102223600 40.42682
## 2018-01-03 43.1325 43.6375 42.9900 43.0575 118071600 40.41978
## 2018-01-04 43.1350 43.3675 43.0200 43.2575  89738400 40.60754
## 2018-01-05 43.3600 43.8425 43.2625 43.7500  94640000 41.06986
## 2018-01-08 43.5875 43.9025 43.4825 43.5875  82271200 40.91732
## 2018-01-09 43.6375 43.7650 43.3525 43.5825  86336000 40.91262

summary(AAPL)

##      Index                 Open             High             Low             Close       
##  Min.   :2018-01-02   Min.   : 35.99   Min.   : 36.43   Min.   : 35.50   Min.   : 35.55  
##  1st Qu.:2019-05-10   1st Qu.: 51.97   1st Qu.: 52.32   1st Qu.: 51.67   1st Qu.: 52.03  
##  Median :2020-09-15   Median :114.67   Median :116.07   Median :112.84   Median :114.97  
##  Mean   :2020-09-14   Mean   :102.38   Mean   :103.62   Mean   :101.24   Mean   :102.48  
##  3rd Qu.:2022-01-20   3rd Qu.:146.36   3rd Qu.:148.00   3rd Qu.:145.15   3rd Qu.:146.61  
##  Max.   :2023-05-31   Max.   :182.63   Max.   :182.94   Max.   :179.12   Max.   :182.01  
##      Volume             Adjusted     
##  Min.   : 35195900   Min.   : 33.87  
##  1st Qu.: 76144000   1st Qu.: 49.73  
##  Median : 98135650   Median :111.96  
##  Mean   :112825725   Mean   :100.02  
##  3rd Qu.:133535000   3rd Qu.:143.97  
##  Max.   :426510000   Max.   :178.65

KDJ指标计算原理

KDJ指标由三条曲线组成：K线、D线和J线。其计算基于以下步骤：

MACD指标择时交易策略分析

量化投资

引言

技术分析是金融市场中常用的分析方法，其中MACD指标是一种重要的趋势跟踪指标，能够反映价格波动的趋势和动量。本研究旨在通过R语言实现基于MACD指标的股票择时交易策略，并通过历史数据回测寻找最佳参数组合。

研究方法

数据获取与处理

我们将使用quantmod包获取股票数据，并使用quantstrat包进行策略回测。首先加载所需的包：

# 加载必要的包
library(quantmod)
library(quantstrat)
library(eTTR)
library(PerformanceAnalytics)
library(ggplot2)
library(dplyr)
library(tibble)
library(scales)
library(gridExtra)
library(showtext)
font_add("SimHei", regular = "SimHei.ttf")
showtext_auto()

接下来，我们获取TSLA的历史数据作为研究对象：

# 设置获取数据的起始和结束日期
startdate.st <- as.Date("2018-01-01")
enddate.st <- as.Date("2023-06-01")

# 获取上证指数数据
getSymbols("TSLA", 
           src = "yahoo", 
           from = startdate.st, 
           to = enddate.st)

## [1] "TSLA"

colnames(TSLA) <- c("Open", "High", "Low", 
                    "Close", "Volume", "Adjusted")

# 查看数据结构
head(TSLA)

##                Open     High      Low    Close    Volume Adjusted
## 2018-01-02 20.80000 21.47400 20.73333 21.36867  65283000 21.36867
## 2018-01-03 21.40000 21.68333 21.03667 21.15000  67822500 21.15000
## 2018-01-04 20.85800 21.23667 20.37867 20.97467 149194500 20.97467
## 2018-01-05 21.10800 21.14933 20.80000 21.10533  68868000 21.10533
## 2018-01-08 21.06667 22.46800 21.03333 22.42733 147891000 22.42733
## 2018-01-09 22.34400 22.58667 21.82667 22.24600 107199000 22.24600

summary(TSLA)

##      Index                 Open             High             Low             Close       
##  Min.   :2018-01-02   Min.   : 12.07   Min.   : 12.45   Min.   : 11.80   Min.   : 11.93  
##  1st Qu.:2019-05-10   1st Qu.: 21.42   1st Qu.: 21.82   1st Qu.: 21.04   1st Qu.: 21.50  
##  Median :2020-09-15   Median :126.48   Median :130.84   Median :122.50   Median :125.58  
##  Mean   :2020-09-14   Mean   :135.13   Mean   :138.26   Mean   :131.71   Mean   :135.07  
##  3rd Qu.:2022-01-20   3rd Qu.:230.08   3rd Qu.:235.64   3rd Qu.:224.40   3rd Qu.:231.07  
##  Max.   :2023-05-31   Max.   :411.47   Max.   :414.50   Max.   :405.67   Max.   :409.97  
##      Volume             Adjusted     
##  Min.   : 29401800   Min.   : 11.93  
##  1st Qu.: 77479900   1st Qu.: 21.50  
##  Median :105881850   Median :125.58  
##  Mean   :134180816   Mean   :135.07  
##  3rd Qu.:159549600   3rd Qu.:231.07  
##  Max.   :914082000   Max.   :409.97

MACD指标计算原理

MACD指标由三条曲线组成：DIF线、DEA线和MACD柱状图。其计算基于以下步骤：

均值回归策略有效性分析

R语言, 金融数据分析

引言

均值回归是金融市场中的一种重要现象，指资产价格或收益率在长期内趋向于回归其历史平 均值。基于这一理论的交易策略通常假设价格偏离其均值后会回归，因此可以通过买入低价 资产、卖出高价资产来获利。

本文将通过R语言实现均值回归策略，并验证其有效性。我们将选取多只股票作为样本，优化 策略参数，并在样本外数据上验证策略的表现。

数据准备与分析

首先加载必要的R包并获取股票数据。我们将选取几只具有代表性的美国科技股作为研究对象。

# 加载必要的R包
library(quantmod)
library(PerformanceAnalytics)
library(foreach)
library(doParallel)
library(ggplot2)
library(dplyr)
library(tidyr)
library(eTTR)
library(zoo)

接下来，我们获取多只股票的历史数据。这里选择了苹果(AAPL)、微软(MSFT)、谷歌(GOOG)、 亚马逊(AMZN)和特斯拉(TSLA)作为样本。

# 设置起止日期
start_date <- "2018-01-01"
end_date <- "2023-01-01"
out_of_sample_date <- "2023-01-02"
end_oos_date <- "2023-12-31"

# 股票代码列表
stock_symbols <- c("AAPL", "MSFT", "GOOG", "AMZN", "TSLA")

# 获取股票数据
stock_data <- list()
for (symbol in stock_symbols) {
   data_raw <- getSymbols(symbol, from = start_date, to = end_date, auto.assign = FALSE)
   colnames(data_raw) <- c("Open", "High", "Low", "Close", "Volume", "Adjusted")
   stock_data[[symbol]] <- data_raw
}

# 获取样本外数据
oos_data <- list()
for (symbol in stock_symbols) {
   data_raw <- getSymbols(symbol, from = out_of_sample_date, to = end_oos_date, auto.assign = FALSE)
   colnames(data_raw) <- c("Open", "High", "Low", "Close", "Volume", "Adjusted")
   oos_data[[symbol]] <- data_raw
}

让我们可视化这些股票的价格走势，以便对数据有一个直观的了解。

基于动量轮动策略的有效性分析

量化投资

引言

动量效应是金融市场中一种重要的现象，指过去表现较好的资产在未来短期内往往继续表现 较好，而过去表现较差的资产则继续表现较差。基于动量效应的交易策略已经被广泛研究和 应用。

在股票市场中，大盘股和小盘股通常具有不同的风险收益特征和市场表现。本研究旨在探索 一种基于动量的轮动策略，通过比较大盘股和小盘股的相对强度，动态调整投资组合的配置， 以获取超额收益。

我们将使用R语言实现这一策略，并通过历史数据验证其有效性。同时，我们会优化策略参数， 测试参数敏感性，并在样本外数据上验证策略的稳健性。

数据准备与分析

首先加载必要的R包并获取股票数据。我们将选取多只大盘股和小盘股作为研究对象。

# 加载必要的R包
library(quantmod)
library(PerformanceAnalytics)
library(foreach)
library(doParallel)
library(ggplot2)
library(dplyr)
library(tidyr)
library(caret)
library(magrittr)

接下来，我们获取股票数据。我们将选择10只大盘股和10只小盘股作为样本。大盘股选取标 普500指数成分股中市值最大的10只，小盘股选取罗素2000指数成分股中市值最小的10只。

# 设置起止日期
start_date <- "2018-01-01"
end_date <- "2023-01-01"
out_of_sample_date <- "2023-01-02"
end_oos_date <- "2023-12-31"

# 大盘股列表
large_cap_symbols <- c("AAPL", "MSFT", "AMZN", "TSLA")

# 小盘股列表
small_cap_symbols <- c("ARQT", "AVXL", "BPMC", "CELZ")

# 所有股票代码
all_symbols <- c(large_cap_symbols, small_cap_symbols)

# 获取股票数据
stock_data <- list()
for (symbol in all_symbols) {
  tryCatch(
    {
      stock_data_raw <- getSymbols(symbol, 
                                   from = start_date, 
                                   to = end_date, 
                                   auto.assign = FALSE)
      colnames(stock_data_raw) <- c("Open", 
                                    "High", 
                                    "Low", 
                                    "Close", 
                                    "Volume", 
                                    "Adjusted")
      stock_data[[symbol]] <- stock_data_raw
      cat("Successfully downloaded", symbol, "\n")
    },
    error = function(e) {
      cat("Error downloading", symbol, ":", conditionMessage(e), "\n")
    }
  )
}

## Successfully downloaded AAPL 
## Successfully downloaded MSFT 
## Successfully downloaded AMZN 
## Successfully downloaded TSLA 
## Successfully downloaded ARQT 
## Successfully downloaded AVXL 
## Successfully downloaded BPMC 
## Successfully downloaded CELZ

# 过滤掉下载失败的股票
valid_symbols <- names(stock_data)
large_cap_symbols <- large_cap_symbols[large_cap_symbols %in% valid_symbols]
small_cap_symbols <- small_cap_symbols[small_cap_symbols %in% valid_symbols]

# 获取样本外数据
oos_data <- list()
for (symbol in valid_symbols) {
  tryCatch(
    {
      oss_data_raw <- getSymbols(symbol, 
                                 from = out_of_sample_date, 
                                 to = end_oos_date, 
                                 auto.assign = FALSE)
      colnames(oss_data_raw) <- c("Open", 
                                  "High", 
                                  "Low", 
                                  "Close", 
                                  "Volume", 
                                  "Adjusted")
      oos_data[[symbol]] <- oss_data_raw
      cat("Successfully downloaded OOS data for", symbol, "\n")
    },
    error = function(e) {
      cat("Error downloading OOS data for", symbol, ":", conditionMessage(e), "\n")
    }
  )
}

## Successfully downloaded OOS data for AAPL 
## Successfully downloaded OOS data for MSFT 
## Successfully downloaded OOS data for AMZN 
## Successfully downloaded OOS data for TSLA 
## Successfully downloaded OOS data for ARQT 
## Successfully downloaded OOS data for AVXL 
## Successfully downloaded OOS data for BPMC 
## Successfully downloaded OOS data for CELZ

让我们计算并可视化大盘股和小盘股的平均价格走势，以便对数据有一个直观的了解。