VIX Historical Survey
Method:
In the period January, 1986, thru September, 2001, there has been a wide range of VIX values (9-150, using the close). Furthermore, there was a period from early 1991 thru late 1997 where volatility remained low with the VIX never touching 30. This study calculated the VIX standard deviation for the previous 50 trading days. A signal was generated whenever the current VIX was greater than the VIX 50 day moving average plus three (3) standard deviations. When multiple signals were generated consecutively, only the first signal was used.
Findings:
In the period March, 1986, thru September, 2001, there have been 32 signals generated. The distribution is as follows:
Number of days
from signal to Number of
next cycle low Occurrances
--------------- ------------
0 15
1 7
2 3
6 3
7 1
9 1
If one bought the S&P on the day the signal was generated and held for three days, the average gain was 0.57%; for 8 days 0.82%; for 30 days 1.96%. The major out-lyer in this survey occurred in October, 1987, when the VIX signal triggered on 10/16/1987 with a value of 36.37. The next day the VIX was 150.19. The S&P low was reached 6 days after the signal. In this case, a 3-day purchase made on the day of the initial signal lost 8.6%; an 8-day purchase lost 17.51%; and a 30 day purchase lost 14.98%.
Using the above criteria, signals were generated on 9/6/2001 and 9/7/2001.
leo |
