# SPY Higher Close Over Next 5 Days, from Alston Mabry

September 30, 2014 |

A study by Kora Reddy showed that for a sample of SPY days:

"4th trading day from the last trading day of the month (i.e 25th September 2014 in this month's case) is down and it is quarter ending (in the months of Mar, Jun, Sep, Dec)"

There were 45 instances, and all 45 had a higher close during the following 5 trading days.

For analysis, starting with all SPY days since 1993, we want to avoid overlap, so we pick every 6th SPY close and determine whether or not it had a higher close over the next 5 days.

Original study:

count: 45
with higher close next 5 tdays: 45
success rate: 100%

Non-overlapping SPY closes (every 6th):
count: 908
with higher close next 5 tdays: 863
success rate: 95%

To get a measure of the significance of the original results, we take our "every 6th SPY close" data set and assign a value of "1" for instances that have a higher high over the next 5 trading days, and "0" for instances that don't. Then we take that series of 1's and 0's and randomly pull 45 observations at a time (with replacement) for 1000 iterations.

Random runs: 1000
Mean sum (of the 45 values pulled): 42.77
SD: 1.437
z of original results (45): +1.55

So if we randomly pull sets of 45 non-overlapping SPY days, we would expect about 43 of them to be successes, i.e., have a higher high over the next 5 trading days. The original results have a z of +1.55 against our randomized runs, so they fall under the common +2 level of significance but are still positive.

To take the analysis one step further, we look at the size of the move from each SPY close:

Original study:
count: 45
mean move to 5-day High: 1.83%
SD: 1.15%

Non-overlapping SPY closes (every 6th):
count: 908
mean move to 5-day High: 1.65%
SD: 1.52%

z of original results (45): +0.81

So, analyzing the size of the Close-High moves shows that, while still positive, the z of the original results is much lower than that obtained by looking at the results as binary, hit/miss observations.

`SELECT * FROM wp_comments WHERE comment_post_ID = '9698' AND comment_approved = '1' ORDER BY comment_date`