Thursday, July 16, 2009

VIX Exhaustion?

More than a few traders noted the weird behavior of yesterday's VIX, which actually rose 3.5%, the same % gain displayed by many of the major averages DJ30, SPX, NDX and IWM.
I'll leave it to more VIX savvy bloggers than myself to parse the reasons for this oddity, but the fact that it happened at all is a reason to rethink some of the underlying assumptions about the VIX's contrary predictive abilities.
Leaving that statistical anomaly aside for a bit , the upper chart shows the current position of the daily VIX bars relative to the LR30 channel.
I've highlighted with white circles the reversal pattern that has developed each time the VIX has displayed a similar channel kiss since May.
Whether history will repeat itself by the end of the week remains to be seen.
If I were a betting man I'd say VIX 28 (LR30 channel mean) has pretty good odds.
An alternate scenario is a channel kiss-off to new VIX lows for the year and the formation of a new LR30 channel with a channel mean in the low 20s.
The only technical clue I see in support of that possibility is the brevity of the latest VIX support/resistance/support cycle, which has only lasted 10 trading days.
The previous cycle was 15 days; the one before that 15 days.
BTW, with today's gains the Qs weekly Pivot Impulse Indicator is now flashing an overbought 1.5 reading.

1 comment:

Discussion Leader... said...

I visit your site every day, and I'm especially glad I did so today--I guess "old farts" perseverate over details too much in this goofy market.

Anyway, I use a prop VIX oscillator that uses variance estimates in combination with a fixed period moving average. The osc has been running very stable since the first of year, with maximum and minimum cycles that are "callable". The 14th I made a post on my blog that "something's up with the VIX." Rather than peaking at +/- 2.0 and then reversing, it made an incursion into the -3.3 region on 07/07. It subsequently reversed and on 07/14 peaked at around +2.7. Those kind of values have not been observed since 10/10/08 to the end of '08, and I've pretty much been baffled at both the RSI and VIX osc action because those two really work well together.

I think your suggestion about the period of the VIX cycle is important here and I will look at that over the weekend. Since a fixed period moving average of data is used in my osc, I'll check to see if the length of the H/L cycle affects the maximum and minimum values.

Cheers and keep up the great work.