Volmageddon: How XIV and SVXY Collapsed on February 5, 2018

Volmageddon is the name traders gave to February 5, 2018, the day the Cboe Volatility Index (VIX) more than doubled and the two largest products betting on calm markets, the VelocityShares Daily Inverse VIX Short-Term ETN (XIV) and the ProShares Short VIX Short-Term Futures ETF (SVXY), lost about 96% of their value between Friday’s close and Monday evening. The losses came mostly in a 15-minute window after the stock market closed, and they were made worse by the products’ own mechanics: to stay on target, both had to buy VIX futures into a rising market, which pushed futures higher, which forced more buying.

In short: XIV and SVXY promised the inverse of the daily return on an index of short-dated VIX futures. On February 5, 2018, that index rose about 96%, so a -1x product had to lose about 96%. Daily rebalancing meant inverse and leveraged VIX products all had to buy futures at the same moment, in a market they dominated. XIV’s prospectus let Credit Suisse terminate the note if its value fell 80% in a day, and Credit Suisse did so the next morning. SVXY survived and cut its target to -0.5x. The episode is the clearest modern case study of how a crowded, mechanically rebalanced derivatives product can destabilize the market it trades in.

This is the institutional companion to implied volatility explained. It assumes you know what the VIX measures. Everything below describes mechanisms and evidence. None of it is a recommendation to buy, sell, or avoid any product.

The misconception: “these products tracked the VIX”

Many XIV and SVXY buyers believed they were short the VIX. They were not. Neither product could hold the VIX, because the VIX is an index calculated from S&P 500 option prices, not something you can own. Both tracked the S&P 500 VIX Short-Term Futures Index ER, a rolling position in the first- and second-month VIX futures, and both delivered the inverse of that index’s return for a single day only.

The SEC said this plainly in a 2021 settled order against S&P Dow Jones Indices (Release No. 33-10943): according to XIV’s pricing supplement, “XIV was not linked to the VIX, but rather to the daily performance of the Index.” The difference mattered for two reasons. First, VIX futures usually trade above the spot VIX when markets are calm (the curve is in contango), so a short futures position earns a “roll yield” as futures drift down toward spot. That drift is what made the products grow. Second, the daily reset is what created the rebalancing flows that broke them.

How the products were built

XIV: an exchange-traded note

XIV was an exchange-traded note (ETN): an unsecured debt obligation of Credit Suisse AG, paying the inverse daily index return less fees. Credit Suisse did not have to hold anything specific; it hedged its obligation by trading VIX futures, but holders owned only a promise from the bank. According to Credit Suisse’s pricing supplement No. VLS ETN-1/A48 (the one in force in early 2018), the Daily Investor Fee Factor for the inverse ETNs was 1.35% a year.

The same supplement contained three sentences that matter for everything that follows:

An Acceleration Event includes “any event that adversely affects our ability to hedge or our rights in connection with the ETNs, including, but not limited to, if the Intraday Indicative Value is equal to or less than 20% of the prior day’s Closing Indicative Value.”
“If the price of the underlying futures contracts increases by more than 80% in a day, it is extremely likely that the Inverse ETNs will depreciate to an Intraday Indicative Value or Closing Indicative Value equal to or less than 20% of the prior day’s Closing Indicative Value.”
“The long term expected value of your ETNs is zero.”

In other words, the termination scenario was disclosed in the issuer’s own words.

SVXY: an exchange-traded fund

SVXY was an exchange-traded fund (technically a commodity pool in ProShares Trust II) that held short VIX futures positions directly and reset its exposure each day to -1x its net asset value (NAV). It had no acceleration clause, so it could not be called away, but it had the same daily arithmetic.

Augustin, Cheng and Van den Bergen (“Volmageddon and the Failure of Short Volatility Products,” Financial Analysts Journal, 2021) put SVXY’s assets at about $1.7 billion and XIV’s at about $1.9 billion in January 2018, with combined assets of about $3.5 billion by early February. They note both products had returned more than 1,000% from inception to peak. ProShares Trust II’s 2018 annual report shows SVXY’s NAV rose 179.0% in 2017 alone.

The day itself

According to the SEC’s order, on February 5, 2018 the Dow Jones Industrial Average fell more than 1,175 points, then its largest intraday point drop, and the S&P 500 fell more than 4%. The VIX “experienced its largest daily increase on record,” closing at 37.32 against Friday’s 17.31 (Cboe closing values), a rise of 115.6%. The Bank for International Settlements (BIS Quarterly Review, March 2018) called it the largest daily increase in the VIX since the 1987 crash.

The timing is what made the day lethal:

Time (ET)	What happened
9:30 a.m. to 4:00 p.m.	Stocks fall; VIX futures rise. Augustin et al. estimate the futures index was up about 39% by 4:00 p.m.
4:00 p.m.	Stock market and ETP trading sessions close. XIV traded at about $99, per the SEC, down from about $110 at the open.
4:00 to 4:15 p.m.	VIX futures keep trading. Front-month futures jump sharply.
4:08 p.m.	Per the BIS, 115,862 VIX futures contracts traded in a single minute, “roughly one quarter of the entire market.”
4:15 p.m.	VIX futures settle. This is the price that sets XIV’s closing indicative value and SVXY’s NAV.
5:09 p.m.	XIV’s closing indicative value published: $4.2217, against $108.3681 on Friday.
About 6:30 p.m.	XIV’s after-hours price reaches a low of about $10, per the SEC.

The S&P 500 VIX Short-Term Futures Index closed at 96.94, per the SEC order. XIV’s value fell 96.1% (4.2217 / 108.3681 - 1). Since XIV paid the inverse of the index’s daily return, that implies the index rose about 96% on the day, from roughly 49.4, which matches the prospectus warning almost exactly: a move well past 80%.

The mechanism: why both long and short products had to buy

This is the part that professionals study, because it generalizes beyond VIX products.

A fund that targets a fixed multiple L of an index’s daily return must rebalance each evening. Suppose it starts the day with assets A and exposure L x A. If the index returns r, the fund’s assets become A x (1 + L x r) and its exposure drifts to L x A x (1 + r). To reset to L times its new assets, it must trade:

Rebalance trade = L x (L - 1) x r x A

The sign of L x (L - 1) is positive for any L above 1 (leveraged long) and for any L below 0 (inverse). So when the index rises, both leveraged long and inverse funds must buy, and when it falls, both must sell. They always trade in the direction of the day’s move, which is the definition of a positive-feedback trader. The BIS made exactly this point: “both long and short volatility ETPs had to buy VIX futures.”

Worked example: a $1 billion inverse product

Take a hypothetical -1x product with $1 billion of assets, short $1 billion of VIX futures at the start of the day.

Index move (r)	New assets	Short futures before rebalancing	Target short	Futures to buy	Buy as a multiple of new assets
+10%	$900m	$1,100m	$900m	$200m	0.2x
+39% (Feb 5 at 4:00 p.m.)	$610m	$1,390m	$610m	$780m	1.3x
+96% (Feb 5 at 4:15 p.m.)	$40m	$1,960m	$40m	$1,920m	48x

The arithmetic has a vicious feature. As the move gets larger, the product’s own assets shrink while the hedge it must unwind grows, so the trade becomes enormous relative to what is left. At +96%, a $1 billion product has $40 million left and must buy back nearly $2 billion of futures. The buying itself raises the price, which raises r, which raises the required trade.

The same formula gives the trade per $1 of assets for different targets:

Target multiple L	L x (L - 1)	Trade per $1bn at +39%	Trade per $1bn at +96%
-1x (old SVXY, XIV)	2.00	buy $780m	buy $1,920m
-0.5x (SVXY after February 2018)	0.75	buy $293m	buy $720m
+1x	0	none	none
+1.5x (UVXY after February 2018)	0.75	buy $293m	buy $720m
+2x (old UVXY, TVIX)	2.00	buy $780m	buy $1,920m

An unlevered long fund (L = 1) never has to rebalance, which is why the problem belongs specifically to leveraged and inverse designs.

The estimates for February 5

Augustin, Cheng and Van den Bergen estimate the scale. Starting from about $3.5 billion of combined XIV and SVXY assets, a 39% rise in the futures index by 4:00 p.m. took the products’ short exposure to about $4.8 billion while their assets fell to about $2.2 billion, so they needed to buy back roughly $2.6 billion of exposure. At the 4:00 p.m. price of the March VIX future (27.95, with a $1,000 multiplier, so $27,950 a contract), that is about 93,000 contracts, which they compare with average daily volume of about 400,000 and open interest of about 600,000 contracts: roughly 23% of a day’s volume and 16% of all contracts outstanding. Adding the leveraged long products TVIX and UVXY, which needed about $550 million more, they estimate about 113,000 contracts had to be bought during the rebalancing window.

The investor class action against Credit Suisse alleges a similar number from the issuer’s side: as summarized by the US District Court for the Southern District of New York in a March 16, 2023 order, the complaint alleges that Credit Suisse “purchas[ed] over 105,000 VIX futures contracts after the market closed,” roughly one-fourth of the VIX futures market. That is an allegation, not a finding.

The authors stress that market participants could anticipate these flows, because the products’ rules were public and the day’s move was visible. They also report that the mid-term inverse ETN (ZIV), which tracked four- to seven-month futures, lost only 1.6% from 8:53 a.m. to 4:00 p.m. and 5.9% through 7:34 p.m. The damage was concentrated where the products concentrated: the front of the curve.

The index that froze

A second, unrelated failure made the evening worse for anyone trading XIV after 4:00 p.m. According to the SEC’s order against S&P Dow Jones Indices, the index’s real-time calculation hit an undisclosed quality-control feature called an Auto Hold, which repeats the previous value when a tick breaches a threshold. The index froze at 87.51 from 4:13:00 p.m. to 4:35:45 p.m., and its 4:15 p.m. closing value turned out to be 96.94. During that window XIV’s published intraday indicative value sat at 24.8933, about 23% of Friday’s close, and it “never dropped below 24, let alone reached the Closing Indicative Value of 4.2217.”

So investors trading XIV after the close saw an indicative value that implied the note had not yet crossed the 80% loss threshold, when its true economic value already had. The SEC found S&P DJI violated Section 17(a)(3) of the Securities Act, and S&P DJI paid a $9 million penalty without admitting or denying the findings (May 17, 2021).

Termination: what the acceleration clause did

At 4:00 a.m. on February 6, Nasdaq halted XIV. Credit Suisse’s press release that morning (filed with the SEC on Form 6-K) said that “an acceleration event has occurred” because the intraday indicative value on February 5 was at or below 20% of the prior day’s closing indicative value. Credit Suisse stopped issuing new notes, set the accelerated valuation date for no later than February 15, and set the acceleration date three business days after that. A follow-up 6-K confirmed February 15 as the valuation date and February 21 as the payment date, and said Nasdaq would suspend trading after the close on February 15.

Holders received the closing indicative value on the valuation date. Per the 2023 district court order, Credit Suisse ultimately redeemed the notes at $5.99 each, about 94% below the $108.37 closing indicative value of February 2.

The litigation

Investors sued. In Set Capital LLC v. Credit Suisse Group AG, 996 F.3d 64 (2d Cir. April 27, 2021), the Second Circuit revived claims that Credit Suisse’s hedging after the close was manipulative and that the offering documents contained actionable misstatements, while agreeing that the complaint did not support scienter for the failure to correct the frozen value. In March 2023 the district court (Judge Analisa Torres) certified a Securities Act class and denied certification of the manipulation and misrepresentation classes without prejudice. In February 2025 it certified a manipulation class, while again denying the misrepresentation class. Plaintiffs estimate that Credit Suisse made between $475 million and $542 million by redeeming at $5.99; that is their estimate, not a finding. As of November 2026 we found no reported judgment or settlement.

The legal question is important for professionals. An issuer that hedges a product must trade, and its hedging can move the price that determines what it owes. The Second Circuit’s decision means that, at least at the pleading stage, hedging activity can support a manipulation claim when the issuer allegedly knew its trades would drive the product’s value down.

SVXY: survival and the -0.5x reset

SVXY had no termination trigger. ProShares Trust II’s 2018 annual report shows SVXY’s NAV reached its 2018 high on January 11 at $552.74 and its low on February 5 at $15.85 (both figures adjusted for a later 1-for-4 reverse split), a fall of 97% from the January peak. ProShares Trust II’s 2019 prospectus lists SVXY’s worst monthly loss as -89.59%, in February 2018.

On February 26, 2018, ProShare Capital Management announced (in a press release filed as Exhibit 99.1 to a Form 8-K) that, effective at the close of business on February 27, 2018, SVXY would seek one-half the inverse (-0.5x) of the index’s daily return instead of -1x, and that the Ultra VIX fund (UVXY) would move from 2x to 1.5x. The release gave no reason. The arithmetic above suggests one. A -0.5x product loses about 48% on a +96% index day instead of 96%, and its required rebalance per dollar of assets falls from 2.0 to 0.75 times the index move, a cut of 62.5%. Halving the target more than halves the feedback.

What regulators and exchanges said

The BIS (Sushko and Turner, Box A in the March 2018 Quarterly Review) concluded that rebalancing by the issuers of leveraged and inverse volatility ETPs appears to have been a key factor behind the volatility spike, and its press release noted that complex products can amplify market jumps “even if the core players themselves are relatively small.” It also documented the market’s growth: average VIX futures open interest went from about 180,000 contracts in 2011 to 590,000 in 2017, and select leveraged and inverse volatility ETPs had about $4 billion in assets at the end of 2017.
The SEC pursued the index-calculation failure (the S&P DJI order above). In October 2020, Chairman Clayton and the heads of three SEC divisions issued a joint statement on complex financial products and retail investors.
Cboe later moved the daily settlement time for VIX futures from 3:15 p.m. to 3:00 p.m. Central (4:00 p.m. Eastern), effective October 26, 2020, per its exchange notices, aligning futures settlement with the equity close. Cboe’s filing tied the change to CME’s own settlement-time change, so it should not be read as a direct response to February 2018, but it does remove the after-close window in which the 2018 rebalancing happened. Since January 2021 the daily settlement price has been a volume-weighted average of trades just before that time; from September 9, 2024 the window is the final 60 seconds.

A separate debate concerns whether the VIX itself can be manipulated through the S&P 500 options used to calculate it at monthly settlement (Griffin and Shams, “Manipulation in the VIX?”, Review of Financial Studies, 2018). That question is about the monthly settlement process, not the daily rebalancing mechanism of February 5, and the two should not be conflated.

How professionals read the episode

Crowding is measurable

Augustin et al. argue that the share of a market held by mechanically rebalanced products is itself a risk indicator. When the products’ required trade on a plausible shock is a large fraction of daily volume, the shock will be amplified. Risk managers now routinely estimate the “rebalance demand” of leveraged ETPs in a given market and compare it with liquidity. The authors explicitly compare the episode to the portfolio insurance losses of 1987: a strategy that sells as prices fall (or here, buys as volatility rises) is benign when small and destabilizing when large.

Short volatility is short gamma

An inverse VIX product is economically a short position in volatility with a daily reset, which behaves like a short option position that is re-hedged every evening. Small daily moves in the futures earn a steady return; a large move produces a loss that grows faster than the move, exactly the profile described in gamma explained. The professional approach is to size short-volatility exposure to survive the jump, not the average day.

Read the termination clause first

XIV’s 80% trigger was the single most important term in the document. Professionals treat termination, acceleration, and “event” clauses as the real product specification, because they define what happens in exactly the scenario that matters. A product that is economically sound on average can still end at the worst possible moment if the issuer has an option to terminate it.

Issuer hedging is a market participant

An ETN holder’s counterparty is a bank that hedges. When the hedge trades in a thin market, the issuer’s own trading feeds back into the index that sets its liability. Both the BIS analysis and the litigation turn on this point.

What can go wrong: lessons that still apply

Daily-reset products are not buy-and-hold instruments. The XIV supplement said the long-term expected value was zero. ProShares’ own release says the funds are “intended for short-term use.”
A single day can exceed the product’s capacity. A -1x product cannot survive a +100% day. A -0.5x product loses half its value on that same day. Any product with leverage or inverse exposure has a move size at which it is wiped out; the professional question is how likely that move is, not whether it is possible.
The settlement window can differ from the trading window. XIV stopped trading at 4:00 p.m. but its value was set at 4:15 p.m. Anyone pricing such a product must know which clock determines value.
Indicative values can be wrong. The Auto Hold showed that a published value can be stale exactly when it matters.
Concentrated flows are predictable flows. When everyone knows a large buyer must buy at a fixed time, others can trade ahead of it, which moves the price further.

The short version

On February 5, 2018, the VIX rose 115.6% and the short-term VIX futures index rose about 96%. XIV and SVXY, which promised the inverse daily return on that index, lost about 96% of their value, most of it after the stock market closed and before VIX futures settled at 4:15 p.m. Because daily-reset leveraged and inverse products must both buy futures when futures rise, their combined rebalancing (estimated by Augustin, Cheng and Van den Bergen at about 113,000 contracts) amplified the spike. XIV’s prospectus allowed Credit Suisse to terminate the note after an 80% one-day fall, and it did; holders were paid $5.99 a note. SVXY survived and cut its target to -0.5x, which reduces its rebalancing feedback by 62.5%. The episode is the reference case for crowding risk in mechanically rebalanced products. For the instrument underneath, see the VIX in depth; for why option sellers usually earn a premium and occasionally lose it all at once, see the variance risk premium.