During the period of time that data covers there were only 2 mid-term congressional election periods, both of which produced a rally. In any event, a sample size of 2 is meaningless data chatter. You cannot draw any conclusions from a set up two. You could have two consecutive heads come up vs tails, which hardly means every flip of the coin will be a tail.
Small sample sizes on fundamentally based relationships is okay, but not that small.
