" Historically, the week following September's Triple Witching has a high probability percentage of being a down week. "
People make all sorts of statements on the net, not all of them are correct. I always verify the statement first before I decide what to make of it.
Just for fun I tested S&P500 over 1980-2009. There are 30 Septembers over this period. 20 out of the 30 post-Triple Witching 5-day returns are indeed negative, so it's 67%.
I'm not sure if they even had Triple Witching before 1980, so I did not test further back.
5-day percentage returns (open of day 1 to close of day 5) sorted in descending order:
7.78
2.42
1.95
1.68
0.64
0.58
0.26
0.07
0.02
0.01
-0.12
-0.26
-0.28
-0.33
-0.38
-0.52
-0.56
-1.17
-1.63
-1.69
-1.83
-2.03
-2.13
-2.13
-2.24
-2.44
-3
-3.35
-3.81
-4.35
Average return: -0.628%
Standard deviation (sample): 2.28%
Note that the 5-day period after Triple Witching is usually the same as the calendar week, but not always, due to public holidays and such.