The backtest didn't lie: what the price curve leaves out

1Alex and the money-making machine

Based on real events: XIV's February 2018 collapse and market prices. Alex—a fictional character—wants to short volatility. He does not short XIV; he buys XIV, a note with daily inverse exposure to a VIX-futures index. He has tested a simple rule: if XIV closes above the average of its last 150 closes, including today's, stay invested; below it, move to cash. Place each order at the following open.

He starts on April 3, 2017 with $10,000. He buys, exits and re-enters over the following months. By December the rule marks $16,261: +62.6%. It fluctuates afterward, but on January 31, 2018 it still marks $15,186: +51.9%. Ten months watching that curve are enough to make Alex excited: “This works. What if I double the bet?”

The first chart stops there, before the collapse. This is the curve Alex saw when he decided to raise the stakes.

On February 1, XIV opens at $125.08. Alex decides to double the units he holds with borrowed money. In the calculation, his own equity at that open is $15,129 and he borrows an equal amount. The new size was not in his backtest. Its barrier is immediate: if XIV falls to $62.54, the position's value equals the debt and his own capital is exhausted. XIV need not go to zero to ruin a 2× bet.

On February 2, the unborrowed curve marks $13,976. On Monday the 5th, XIV still closes on the exchange at $99 and that curve marks $11,974. The moving average orders an exit, but that order reaches only the next open. The first February 6 opening price in the exchange-price archive is $10.49. At that price the 1× rule ends at $1,269: −90.9% from February 2. The same quote is far below the 2× own-equity barrier of $62.54. The SEC records that XIV was halted ahead of the usual open: a stop could not hand you the previous close.

The second chart reveals the hit. Above is the same rule after the gap; below is a close-up of February 1–6. The 2× line stops at the own-equity barrier: the next printed price crosses it. A broker might have demanded margin or liquidated earlier; the line does not purport to reconstruct a particular liquidation.

Alex looks at the complete curve. The backtest didn't lie. He asked only what that rule earned on the prices he knew. He had not asked what he owned, who could close the product, or whether his account could survive a gap.

A historical price records what happened. It is not a prospectus, an executable order, or your account agreement. The curve tells a story; the contract decides when it ends, the order what you can sell for, and your account whether you can wait. An out-of-sample window is still history: if neither window contains the event or models the rule you need to test, splitting the windows cannot conjure it into existence. A disaster can appear on the chart afterward. That does not mean the earlier chart explained how it could happen.

2XIV: the document outside the screen

XIV was an exchange-traded note—an ETN—issued by Credit Suisse and linked to the inverse of the daily return of a VIX-futures index, not simply the inverse of the VIX itself. The distinction matters: the index and the note had their own mechanics; this was not a permanent, direct wager against a volatility number. An SEC order concerning an adviser's marketing of XIV and VIXY describes that exposure and the offering warnings.

Credit Suisse announced acceleration on February 6, 2018. According to its release, the February 5 event met the stipulated threshold: intraday indicative value at or below 20% of the previous session's closing indicative value. The release said the payout would refer to the closing indicative value on the accelerated valuation date, not an automatic sale at the last screen price. Trading price, intraday indicative value, closing indicative value and eventual cash payout are different measures answering different questions.

Here is the scene the exchange close did not show. XIV had closed at $99 on the exchange on February 5, but the SEC's index order documents a $4.2217 closing indicative value, calculated at the 4:15 p.m. VIX-futures close and disseminated at 5:09 p.m., New York time. The next day, according to the historical market-price archive, XIV opened at $10.49 and closed at $7.35; the SEC's XIV adviser order records the eventual $5.99-per-note payment. Those figures are not one homogeneous series. The equity chart uses exchange prices; the clause and payment use indicative values at different moments. Mixing them to claim a fill or a return would mislead.

There is a revealing complication: that SEC investigation documented that none of the intraday indicative values published that day showed the 20% threshold crossing, because an undisclosed automated control held some index data static. The closing indicative value did fall below it. That is why we attribute the trigger to the issuer's announcement; we do not present it as a signal any investor could have checked on their screen in real time. Imprudent sizing and missing stress tests deserve criticism; the full damage cannot simply be blamed on someone watching a public feed the SEC found defective.

The clause existed before the episode; the loss appeared afterward. Nor would it be fair to blame every loss on one line of legal text: the index move, daily inverse exposure and product mechanics all belong to this case. The line adds one crucial fact: the investment could end without the holder choosing to end it. A backtest of closing prices does not, on its own, implement that contract.

2.1. What professional review would have asked beforehand

A moving average measures observed prices; it cannot replace a scenario absent from the sample. Before approving a position like this for weeks or months, risk review could have asked three uncomfortable questions: what happens to capital if the note loses 80% or 100% of its indicative value and accelerates? What does a stop actually achieve if damage arrives after hours and the next open has an enormous gap? At what price does a borrowed 2× account run out of its own equity?

The answer is not “predict February 5.” It is to decide beforehand whether the instrument suits the holding period and monitoring capacity, and how much voluntary damage can reach the wider portfolio. One illustrative professional decision would have been to exclude XIV for this multiweek rule, or hold it without borrowing as a small part of a diversified portfolio. For example, with 5% of total capital in an unleveraged note, a complete loss on that note directly damages 5% of the portfolio before correlated moves, costs or other losses. That is neither a published Qinvia policy nor a position-size recommendation: it shows why contract, execution and exposure limits must be reviewed beside the backtest.

3Five ways to test the curve and miss the object

These categories are neither mutually exclusive nor exhaustive. On a bad day, several may meet.

3.1. The contract: what claim do I own, and when does it end?

An ETN looks much like an ETF on a broker's screen. Legally it is different: an unsecured debt obligation of the issuer. You can be right about the index and still bear the credit risk of whoever owes the payment. Default may bring partial or total losses; not every issuer failure automatically reduces the claim to zero. The SEC's ETN bulletin also warns that market price can diverge from indicative value and that payment terms live in the prospectus.

And “the product closes” can mean different things. In an ordinary fund liquidation, the fund's assets are generally distributed to shareholders. There may be costs, taxes, a different proceeds price and a need to reinvest, but it is not an ETN's market-triggered contractual acceleration.

A curve is not a right to payment. Search the document for “acceleration,” “early redemption,” “call,” “issuer credit risk” and the valuation and payment formula. Can the contract end your position while your market thesis is still alive?

3.2. The structure: what happens inside the product between closes?

A quick calculation: an underlying moves from 100 to 110 and back to 100. Two days; unchanged destination. An idealized vehicle targeting 2× each day rises from 100 to 120 and then falls to approximately 98.18. Its two-day return is not “twice zero.” The 10-point rise is measured against 100 (+10%); the 10-point fall is measured against 110 (−9.09%), not 100. It compounds daily factors: 100 × (1 + 2×0.10) × (1 − 2×1/11) = 98.18….

One subtlety: this effect does appear in the actual product's history. You hide it when you replace the product with “twice the underlying's return from start to finish.” Collecting option premiums is not free upside either: income comes with limits on participation, costs and adverse scenarios. What your data leave out depends on the proxy you chose.

“2× today” isn't “2× until I sell.” Read the period over which the product makes its promise—a day, a defined outcome period, or something else—and simulate its actual rules and cash flows rather than a naive multiple of the underlying.

3.3. Liquidity: at what price can I leave?

A backtest selling everything at the closing price assumes that close was an offer available for your order, in your size and at the required moment. An ETF trades at a market price that can be above or below its NAV, and the SEC cautions that a trading market may fail to develop. The gap between the last trade, the midpoint and the best bid can matter most when everyone wants out.

The last price isn't your next sale. Inspect spreads, volume, premium/discount to NAV or indicative value, depth and creation/redemption terms. What would your backtest assume when the order cannot meet a buyer at the displayed price?

3.4. Your account: who may sell on your behalf?

The prospectus is not the only fine print. Buying on margin puts the position inside an agreement with your broker. FINRA explains that firms can sell assets in a margin account without issuing a call first, without letting you choose what to sell and, in some circumstances, sell more than needed for the immediate shortfall. House maintenance requirements can exceed regulatory minima.

The asset's curve might fall and recover. A liquidated account does not automatically recover the units it no longer holds. We assign neither a universal threshold to brokers nor a fabricated margin trade to XIV.

A recovery can't return units already sold. Read the margin agreement, house maintenance requirements and sale-without-notice authority. Would your strategy survive if someone else selected your exit time?

3.5. Operations: what if you cannot act today?

Liquidity asked at what price you could leave when a market is open. The prior question is whether you can even place and execute the order. A trading halt or extraordinary market closure may prevent an exit while it lasts; an outage of your broker access may cut off your usual route even as prices move. Order execution is not instantaneous and does not guarantee the price on your screen. The product prospectus cannot guarantee that your order will reach the market at the moment you need it.

Deciding to sell is not the same as being able to sell. Ask what you would do during a halt or access outage; in a stress test, leave the order unfilled or delay it instead of filling it at the last close.

4The fine print is still alive

These are not inherently “bad” products. Each may address a real need while charging for its features in a way worth understanding.

A single-stock leveraged ETF combines daily reset with the concentration risk of one company's shares. The SEC's investor-education staff warns that holding beyond its objective period can produce a result very different from multiplying the stock's cumulative move.

A defined-outcome ETF, such as the Innovator U.S. Equity Power Buffer ETF — September (PSEP), seeks to buffer the underlying's first losses measured from the start to the end of the outcome period, in exchange for an upside cap; the outcome is subject to fees and not guaranteed. Buying halfway through is not buying at the start: look at the remaining buffer and upside potential from your own purchase price, as well as the current prospectus.

Among option-income funds, JEPI combines equities with call-writing implemented through equity-linked notes (ELNs): it gives up some upside in exchange for income, and the fund bears counterparty and liquidity risk on those notes, according to its prospectus. Owning JEPI shares is not the same as being a direct creditor of XIV's issuing bank.

YMAX holds YieldMax option-strategy funds; its prospectus dated February 2026 and supplemented in June warns about participation limits and distributions that may include a return of capital, with a risk of NAV erosion in its underlying funds. JEPI and YMAX are not the same strategy; neither warrants a blanket claim of inevitable capital erosion.

In a hypothetical example, if a fund distributes 5 and its NAV mechanically declines by 5 on the ex-dividend date, that does not prove an economic loss of 5: the holder has the payment as well as the remaining fund interest. Evaluate total return including distributions, a suitable benchmark, taxes and the nature of payments; a distribution rate alone is not a return earned. For any current ETN, return to the question of who owes the payment, how it is valued and whether early acceleration is possible. The contractual risk need not be illustrated with a new ticker to remain relevant.

5Five questions worth more than another pass through the curve

1. 1Can my claim be permanently impaired or my position terminated? Read “principal risks,” acceleration, liquidation and payment formulas; distinguish market loss, termination and routine fund closure.
2. 2What transforms the underlying path, and who can force my exit? Read daily/defined-period objective, resets, derivatives, financing and the margin agreement.
3. 3Who owes me money, and what if they fail to pay? Distinguish shares in a fund that owns assets from unsecured issuer debt; inspect counterparty exposures where relevant.
4. 4Can I get out, and at what price if everyone sells? Compare close and NAV/indicative value with bid/ask and depth; also consider a halt, delayed order or access outage, not just average volume.
5. 5What does the quoted “yield” mean? Separate premiums, dividends, return of capital, NAV changes and total return after costs; check distribution notices and the prospectus.

This is neither a ban on complex instruments nor personal investment advice. It is a way to ask the complete question. A backtest established how a rule would have behaved on observed prices under programmed assumptions. Alone, it did not establish that the contract, structure, market and your account would let you realize that curve. When a strategy next looks wonderful, keep the chart: put the prospectus and an exit order beside it. The simpler the instrument—without direct issuer debt, internal resets, derivatives or contractual acceleration—the fewer layers to decode; liquidity, access and position size still matter.