Very interesting how its all within 40-60% of what happens next. I don't consider the 6-8 down days in a row to be at all significant because its based on such a small sample size.
It would be interesting if they broke it down even further, like less than and greater than 1%. For example, 3 days of great than 1% down might give a better signal than 3 days of less than 1% down each.