The critical part about severe oversold readings of “eFear” is not that we just get to buy the market with confidence around these points but rather that it takes time for shorts to be covered or positions rebuilt by major participants, after a significant pullback.
We must look at 2 possible scenarios:
1. We buy the market if EFear has ever registered a -9 or lower in the previous N days
2. We buy the market only if it has never been below that level for the past N days
We can see a huge disparity between returns when efear has been below -9 sometime in the last 20 days. Ten days seems optimal but we must see an equity curve to make that choice.
It is clear that market returns continue to skew tremendously in favor of being long for some time after risk has hit significant lows and that this is a dissipating force.
Do Bonds Confirm?
Given the negative correlation between stocks and bonds on a daily basis we should also see comparatively weaker returns for the 10 year bond futures contract (during the same windows when S&P has recently experienced a “fear event”). Over the entire sample period the bond contract rallied 56.34 points.
This is clearly true. After ten days the advantage wanes. Inside ten days you make more (or as much) money shorting bonds than buying them – during a huge long term bull market.
Let’s compare returns for different levels of EFear inside vs outside different memory windows. Here are the rules:
o A time window begins if we cross below the threshold and the 10 day average of EFear is not less than that threshold.
o If the 10 day average subsequently falls below the threshold then the day count is reset to zero (so no data is collected)