Volmageddon, February 2018

Through 2017, short-volatility exchange-traded products had become one of the market's most crowded trades. Funds like XIV and SVXY let ordinary investors take the "short VIX futures" position that had been quietly profitable for years, because implied volatility usually sits above the volatility that actually shows up — a spread traders can collect by selling VIX futures and rolling the position each month. It worked, right up until the day realised volatility itself moved.

What happened

On 5 February 2018, the S&P 500 fell about 4% intraday — a real but not extreme move. What made the day unusual was what it did to the VIX index, which measures expected volatility: VIX jumped from around 17 to over 37, a 116% one-day increase, its largest ever recorded.

Time (ET)	Event
Market close, 2 Feb	S&P down about 2% for the week; VIX elevated but unremarkable
Day session, 5 Feb	S&P falls roughly 4% intraday, a normal-sized selloff
After 4pm settlement	VIX futures spike as dealers hedging short-vol ETP rebalancing needs buy aggressively into a thin, after-hours market
~4:15pm	VIX closes up 116% on the day; short-VIX-futures ETPs down 80-95% after hours
Next morning	Credit Suisse announces it will terminate XIV; the fund is liquidated within days

The mechanism

XIV and similar products promised daily inverse exposure to VIX futures, and to deliver that consistently each day they had to rebalance their futures position every afternoon — buying more VIX futures as VIX rose, to keep their leverage ratio constant (the same mechanical effect that drives End-of-Day Rebalancing Flows of Leveraged ETFs). That rebalancing itself was public knowledge and predictable in size and timing.

As the market fell on 5 February, VIX futures started rising, which meant the short-VIX ETPs needed to buy futures to rebalance — buying into a thin, late-afternoon market, right as issuers were also racing to hedge against the products being wiped out entirely. Each dollar of losses pushed VIX futures higher, which forced more buying, which pushed VIX futures higher still. A predictable, mechanical rebalancing flow, once large enough relative to the market it traded in, became self-reinforcing. XIV lost roughly 96% of its value after hours in under an hour and was terminated by its issuer within the week; investors holding it that evening were wiped out.

The lesson

A strategy that is profitable specifically because it is short a tail risk (here: a volatility spike) will eventually meet that tail, and the loss on that one day can exceed every dollar it ever made. Selling volatility resembles picking up pennies in front of a very occasional but very real steamroller — see VIX vs Realised Volatility Spread Trades. The second lesson is about scale and liquidity: a rebalancing rule that is safe at small size can become the dominant flow in its own market once enough capital follows the identical rule at the identical moment, and predictable flows invite front-running by anyone able to anticipate them, which amplifies rather than dampens the move.