Blogs / Analysis Blog

Event Study: Performance after VIX Spikes

A common type of analysis in finance is looking how events affect the markets.  These can be anything ranging from earnings announcements, split dates, breaking news, or even more complicated technical events.

Palantir allows us to isolate the events–as simple as manually inputting dates or as complicated as multiple technical filters–using Date Set.  These can then be tested in Strategy and then looked at cross-sectionally across different groups using Twiddle.

This study will look at the dates that the VIX spikes upwards as the trigger event, then see how performance of the S&P 500 and its individual sectors are for the in the days following.

Tracking the Event

First we need to isolate all the days when VIX spikes upwards.  To do so, we’ll look at the 63-day Z-Score of the VIX, and choose the dates when that z-score crosses above 2.  Below, we can see that that hasn’t happened yet in 2009 but happened 8 times in 2008, most recently on October 22. 

 image001

Now we use this Date Set as an input to our Strategy.  The Strategy below is a simple Strategy that holds the S&P 500 for 5 days after each VIX spike, starting in 1998. 

image003

The NAV curve is below.  The Strategy shows very mixed, inconsistent results.

 image005

Using Twiddle, we can see a sector-by-sector breakdown of performance.  Instead of only looking at 5 days after the event, we’ll run backtests for 5 days to 60 days across all sectors. and look at Sharpe Ratios.  Columns represent sectors and rows represent number of holding days after the event.

Each sector does well 5 days after a VIX spike, but the strongest performers after that are the Energy Sector, Health Care Sector, Consumer Staples Sector, and Technology Sector.  The weakest performers are the Materials Sector, Utilities Sector, and Consumer Discretionary Sector. 

image007

This gives us a snapshot of how the S&P and its sectors do after an event.  It’s a framework for how to conduct an event study in Palantir and can be applied to anything from earnings announcements to other technical indicators.

Other Blogs