The ISEE sentiment index is the ratio of opening long call options to opening long short options. The idea is that the greater the ratio of calls, the more bullish the sentiment, and that this is a more reliable indicator (compared to other sentiment indices) because it’s based on actual trades as opposed to surveys. Compared to the AAII sentiment numbers it also has the enormous advantage of being available almost in real time, instead of updating once a week. Keep in mind though that it’s only based on trades on the ISE rather than the entire market.

There are a couple of different versions: one for all securities, one for equities only, and one that’s ETFs only. I will be using the equity ISEE for this post. Can we use it to trade? Let’s start with some charting. Here’s ISEE plotted against SPY:

Visual inspection seems to suggest a contrarian trading strategy, as bottoms in the ISEE tend to correspond to bottoms in SPY. But is there a way to pick out the bottoms and create a profitable trading strategy?

The actual level of the ISEE index is not strongly predictive of returns. Higher levels are associated with slightly higher next-day returns for SPY (so the contrarian intuition was probably off), but the effect is nowhere near large enough to trade on. The same holds for predicting returns over longer periods.

I was also unable to to find a way to pick out good trades based on ISEE n-bar lows, n-bar percentile, and other similar approaches to picking out extreme lows. I also looked into whether bounces after lows were good entry points, but again no luck.

Could we use it for risk management? There’s a slight relation between ISEE extremes (both high and low) and more unexpected future volatility, but it’s not very strong so again I wouldn’t use it.

Sentiment as a Filter?

What about using it as a filter to improve another trading strategy? Here it may have some potential, in a “go with the crowd” type of approach. Here’s a super simple strategy, if RSI(3)<20 at the close, it buys SPY and then sells at the next close. The strategy earns 19bp per day it spends in the market (without taking into account commissions). Filtering out the days when ISEE is below the 20th percentile of its values over the past 50 days improves the returns to 29bp and significantly reduces risk. This also works on the opposite side, as low ISEE values tend to be more favorable for shorts.

But the effect is marginal and given the lack of predictive ability the ISEE has otherwise, I wouldn’t really trust it to work reliably in the future. Overall this is probably not something I would use.