Friday, September 12, 2014

$VIX is >+2SD, except not really


Vix is >+2SD, which is a good time to not be in the market:


Except it's not >+2SD:


It's only >2 standard deviations above norm if you use a period of 20. That's because of that little $VIX spike we had more than 20 days ago: if you select a period of 20, it falls out of the equation, and thus the Bollinger bands for a 20 period tightened a few days ago.

The Bollingers for a 30 day period (second chart) are still big because the spike hasn't fallen out of the dataset yet.

Thus today's $VIX is only a statistical outlier if you cherry-pick your statistical set to make it so.

That's an important thing to remember about all statistical TA indicators: they depend on the dataset. If your local TA doesn't quality-check the dataset for sources of error, then he's an idiot and you should ignore him from now on.

So, because the Bollinger function is telling me two different things depending on the dataset, I'm going to ignore the Bollinger indications for this present situation, and concentrate on $VIX's relation to the short-term EMA (which shows it still in an uptrend) and also keep an eye on the VIX futures curve (not remotely inverted yet at all, still boring) and HYG (which has been doing nothing for 2 days, but still has a chart that hints there might be a change taking hold in the character of the high-yield market's performance).


No comments:

Post a Comment