Wait Times Between Big Drops, from Kim Zussman
SPY lost 2.5% last week. It has been 23 weeks since it dropped more than 2%. From 1993-present, checked for weeks down more than 2%, and the wait time (in weeks) between such drops. Then checked return of the following weeks.
First here are mean return weeks after 2+% drops regardless of wait times:
One-Sample T: nxt wk
Test of mu = 0 vs not = 0
Variable N Mean StDev SE Mean 95% CI T
nxt wk 151 0.0020 0.0400 0.0032 (-0.0044, 0.0084) 0.62
Mean up 0.2% with SD of 4%
Here are mean returns of weeks after 2+% drops, for waits between such
drops longer than 20 weeks:
One-Sample T: nxt>20
Test of mu = 0 vs not = 0
Variable N Mean StDev SE Mean 95% CI T
nxt>20 12 -0.0021 0.0192 0.0055 (-0.0143, 0.0101) -0.38
Down -0.2% with about half the volatility: SD = 1.9%
(volatility clusters).
This is consistent with the sign of slope coefficient for regression of
next week's return vs wait time (though NS):
Regression Analysis: nxt versus wait_1
The regression equation is
nxt = 0.00240 - 0.000071 wait_1
Predictor Coef SE Coef T P
Constant 0.002405 0.004269 0.56 0.574
wait_1 -0.0000706 0.0003814 -0.19 0.853
S = 0.0403364 R-Sq = 0.0% R-Sq(adj) = 0.0%
rather than waste time on analysis it is easier to just wait for Rocky's calls.