Sentimentally Speaking

A pet peeve of mine with many indicators (like Barry’s Traffic Indicator to name one) is that there’s often failure to ask whether it tells you anything you can’t simply infer from just looking at an index. In other words, take Bullish sentiment. If it just simply rises with the market, and then peaks and falls with the market, where’s the value added? It’s only noteworthy if/when it ever diverges from expectations and tells you something you wouldn’t see otherwise.

So with that in mind, Jason Goepfert highlights a study that examines this very phenemenon. And as it turns out, our beloved VIX (shown here in bikini form) stands out.

They take a new (to me, anyway) approach to strip out price influence from a handful of popular sentiment indicators to find whether or not it improves their predictive power. In all cases but one, it did not (suggesting that price is an important component of what makes sentiment tick).

The standout exception is VIX. Figures 27 and 28 show purification produces a strong improvement PF. While only 2 of 10 rules based on unpurified VIX beat a random signal, 9 of 10 rules based on purified VIX display a significant PF. This suggests that VIX contains predictive information above and beyond price dynamics that is masked by the strong influence that price dynamics have on VIX. We believe that purified VIX represents an improvement over standard VIX, and price dynamics purification represents a step forward in sentiment analysis in general as it can point to indicators that contain information that is not redundant of that found in price indicators.

The VIX contained information not wholly redundant with price, and stripping that influence out of the indicator improved its efficacy.

The study is here, but I can’t get the link to work, so will hopefully have a better link later for anyone who wants to take a look.

Now as I take it from Jason’s description, it’s consistent with the relatively amateurish stats analysis I did for my book. The VIX in and of itself is not wildly predictive of anything you wouldn’t see just from watching the SPX. BUT, I did indeed find that VIX behaving off expectations did have implications 5, 15, and 25 days out. In other words, let’s say the market flies and the VIX holds strong, it does suggest outperformance going forward over different time frames.