The
easiest trade in the market is in XIV and SVXY – just buy the dip,
right? Wrong! In very calm markets that may be true. Gains in these
funds are compounded daily as long as volatility is decreasing and
futures are in contango. It really is a mindless trade when things go
well. From January 1, 2017 to its all-time high on July 24, XIV rose
105%, in what was mostly a steady, daily rise with few interruptions.
After
late July though, the behavior on the XIV and SVXY changed. The
overall term structure of the VIX futures is certainly a little higher than it was in July, but both of
these funds are down 17% from their highs in the last 6 weeks, even
though VIX continues to stay in the low teens. What's going on under
the hood?
Beginning
in late July the market began to see frequent VIX spikes, on average
once every 7-10 days. Although these spikes weren't high in absolute
terms, on a percentage basis they were quite large:
Date
|
VIX HIGH
|
% SPIKE FROM LOW
|
---|---|---|
July 27 | 11.50 | 30.1% |
August 11 | 17.28 | 81.4% |
August 18 | 16.04 | 42.6% |
August 29 | 14.34 | 29.2% |
September 5 | 14.06 | 40.3% |
VIX futures don't move nearly as much as the VIX index, and therefore the volatility funds were not exposed to moves of this magnitude. Still, once a fund starts to move in double-digit percentage moves every day, it begins to suffer losses from something it has to do every day - rebalancing.
Because
the short funds SVXY and XIV track the inverse of the daily
percentage
change in VIX futures, it has trouble making up lost ground from a
large drawdown. For example, suppose VIX futures jumped 20% from
12.00 to 14.40, then give it all back on the next day to settle again
at 12.00. If this move happened over the course of two days, SVXY
would see a 20% loss on the first day. On the next day, however, SVXY
would only see a 16.67% increase (a futures drop from 14.40 to 12 is
a 16.67% loss). Therefore, in absolute terms, if SVXY started the
move at 90, it would drop 20% to 72, but only rebound 16.67% to 84,
even though the
net change in futures over the two days was zero.
If
that type of move is repeated once a week, the losses from frequent
drawdowns would rapidly outpace any gains from contango or a broader
move lower in volatility. While this effect may not be easily
apparent since VIX futures are now considerably higher than they were
in July, it helps to do a quick calculation.
As
of the writing of this article, SVXY's current price is 77.70 for a
30 day weighted average future of 13.683. If SVXY were to return to
its all time high of 93.43, it would need to see a 20.2% drop in
futures to 10.92. This would be extremely unlikely, as it didn't even
happen at the all-time VIX low of 8.84, when the 30-day weighted
average future bottomed out at 11.33. An additional 3.8% move lower
would be required.
SVXY
requires
the presence of contango and longer, gradual moves lower in the
futures to compound gains and catch up to where it was before being
hit by the frequent drawdowns of relatively massive VIX spikes.
In
summary, even though SVXY and XIV might be great buy-and-hold
positions in calm markets when VIX spikes are very few and far
between, the results can be devastating in the environment of
rapid-fire VIX spikes, even when VIX is low in absolute terms.
Your diagram of how VIX volatility and rebalancing impacts SVXY is excellent, the clearest explanation I've seen.
ReplyDeleteinteresting math. so how do you reconcile your thoughts with this point of view? thanks.
ReplyDeletehttps://seekingalpha.com/article/1580492-do-1x-inverse-etfs-erode-value-from-volatility-actually-no
I've read that article before and it leaves me confused. He somehow tries to make the argument that if you short something at 100 and cover it 110, you only lost 9% because 10/110 is 9%. I guarantee however, that is a 10% loss no matter how you slice it because if you sold something at 100 and had to buy it back at 110, you spent 10 plus the 100 you made on the sale to buy it back (therefore losing 10%). I'm not sure if that author is just not explaining it well or just doesn't understand how shorting works.
Delete