Poll trends

N_all <- 1000
out <- numeric(N_all)
N <- 50

for(ii in 1:N_all){
  set.seed(ii)
  tmp <- round(rnorm(N, 50, 1.9), 0)
  tmp <- ((tmp[-1] - tmp[-N]))
  
  out[ii] <- sum(zoo::rollapply(tmp, 3, function(x) all(sign(x)==sign(x)[1] )))
  
}

library(ggplot2)
p1 <- ggplot(data.frame(x=out), aes(x=x)) + geom_bar() +
  theme_minimal() + ggtitle("3 Poll Trend") + 
  labs(x="Number of Instances in 50 Polls") + 
  scale_y_continuous("Percentage", 
                     labels=scales::label_percent(accuracy=1, 
                                                  scale=.1))

for(ii in 1:N_all){
  set.seed(ii)
  tmp <- round(rnorm(N, 50, 1.9), 0)
  tmp <- ((tmp[-1] - tmp[-N]))
  
  out[ii] <- sum(zoo::rollapply(tmp, 4, function(x) all(sign(x)==sign(x)[1] )))
  
}
p2 <- ggplot(data.frame(x=out), aes(x=x)) + geom_bar() +
  theme_minimal() + ggtitle("4 Poll Trend") + 
  labs(x="Number of Instances in 50 Polls") + 
  scale_y_continuous("Percentage", 
                     labels=scales::label_percent(accuracy=1, 
                                                  scale=.1))

p <- gridExtra::grid.arrange(p1, p2,top="How Often Do Polling Trends Appear?")
ggsave(p, file="~/Desktop/trends.png", height=8, width=6)