sector-seasonality.R

#Analyze median sector returns by month of the year to find the best-performing sectors for each month

library(tidyverse)
library(tidyquant)
options(digits=2)

ticker <- c("EWRE","RCD","RTM","RYE","RYF","RHS","RYH","RYU","XAR","XBI","XHB","XME","XTL","XTN","XSW","XSD")
start <- as.Date("01-01-1980",format="%m-%d-%Y")
end <- as.Date("12-31-2020",format="%m-%d-%Y")

seasonality_finder <- function(ticker){
  price_data <- tq_get(ticker,from = start,to = end)

  monthly_return <- price_data %>% 
    mutate(year_month = paste(year(date),month(date))) %>% 
    mutate(previous_close = lag(adjusted)) %>%
    mutate(return = c(NA,diff(adjusted))/previous_close) %>%
    group_by(year_month) %>% 
    summarize(return = sum(return)) %>% 
    filter(!is.na(return)) %>%
    separate(year_month,c("year","month")) %>% 
    group_by(month) %>% 
    summarize(return = median(return)) %>%
    mutate(month = as.numeric(month)) %>%
    arrange(month) %>%
    mutate(month_of_year = c("January","February","March","April","May","June","July","August","September","October","November","December"))
  
  monthly_return$month_of_year <- factor(monthly_return$month_of_year,levels = monthly_return$month_of_year[order(monthly_return$month)])
  
  monthly_return <- monthly_return %>% mutate(ticker = ticker,positive = return >0) %>% select(ticker,month,month_of_year,return,positive)
}

tidy_seasonality <- map_dfr(.x=ticker,.f=seasonality_finder)
best_of_month <- tidy_seasonality %>% group_by(month_of_year) %>% summarize(best_ticker = ticker[which.max(return)],return = max(return))

best_of_month %>% ggplot(aes(x=month_of_year,y=return,fill=return)) +
  geom_col() +
  labs(title = "Best-performing (equal-weight) sectors by month of year",x = "Month",y = "Best-performing sector's median return") +
  geom_text(aes(label = best_ticker), vjust = -0.5) +
  scale_y_continuous(labels = scales::percent) +
  theme_classic()

ggsave(filename = "sectorseasonality.jpg",
       path = "C:/Users/chris/OneDrive/Pictures/Infographics",
       scale = 1,
       width = (1920*1.5)/300,
       height = (1080*1.5)/300,
       units = "in")

#Backtest a sector-switching strategy for 2021 to compare YTD returns for sector-switching vs. holding the equal-weight index

switching <- data.frame(month = 1:12,ticker = c("XBI","XAR","EWRE","RYE","XSD","EWRE","XSD","XSD","RTM","XTN","XTN","XME"))
month_of_year = c("January","February","March","April","May","June","July","August","September","October","November","December")
start <- as.Date("12-31-2020",format="%m-%d-%Y")
end <- min(Sys.Date(),as.Date("12-31-2021",format="%m-%d-%Y"))

RSP_data <- tq_get("RSP",from = start,to = end)
RSP_monthly_return <- RSP_data %>% 
  mutate(month = month(date),return = c(NA,diff(adjusted))/lag(adjusted)) %>%
  group_by(month) %>% 
  summarize(return = sum(return)) %>%
  filter(!is.na(return)) %>%
  mutate(month = as.numeric(month)) %>%
  arrange(month) %>%
  mutate(ticker = rep("RSP"),
         strategy = "Index strategy")

return_finder <- function(ticker){
  price_data <- tq_get(ticker,from = start,to = end)  %>% 
    mutate(month = month(date),return = c(NA,diff(adjusted))/lag(adjusted)) %>%
    filter(!is.na(return))
  months_to_use <- switching$month[which(switching$ticker == ticker)]
  price_data <- price_data %>%
    filter(month %in% months_to_use)
  
  monthly_return <- price_data %>% 
    group_by(month) %>% 
    summarize(return = sum(return)) %>%
    filter(!is.na(return)) %>%
    mutate(month = as.numeric(month)) %>%
    arrange(month) %>%
    mutate(ticker = rep(ticker),
           strategy = "Sector switching strategy")
}

tidy_return <- map_dfr(.x=unique(switching$ticker),.f=return_finder)
tidy_return <- bind_rows(tidy_return,RSP_monthly_return)
total_return <- tidy_return %>% group_by(strategy) %>% summarize(sum(return))