I’ve read numerous articles attempting to use the VIX (or a normalized VIX) as an indicator of market bottoms. In fact, CNN introduces it as the first factor in their “Fear and Greed” index. But instead of forecasting and prediction, perhaps we should be looking at it in a different way…
More risk should be compensated by more reward. VIX (which is constructed using the near and next-term options) does a fairly decent job of predicting forward volatility (top right). However, we don’t see a similar pattern in returns (especially in the 18-25 buckets). In fact, the forward returns drop in the 18-25 bucket. This is the case even when removing the 2008 financial crisis from the data set.
1) When things get bad and uncertain, they’re liable to get more bad and uncertain (dumping positions is a valid play).
2) Market participants (in general) are more long than short. Given some level of leverage, higher volatility -> higher required portfolio margin -> more selling. By the time we get to > 25 (~1.6% moves / day), all the speculators have already been shaken out (hence, why the risk/reward normalizes).
3) I’m fooled by randomness (especially given the market crashes in the last decade) and the risk/return is actually consistent across vol buckets.
4) I’m fooled by bad math and faulty statistics.
* Includes ~ 14 years of data, starting in 1/1/1990 (the back-filled VIX data).
* I haven’t looked at it 2nd order (ie, changes in VIX), but something I want to explore.