The VIX & Its Term Structure

You already know realized vol from implied vol, you’ve priced a variance swap as a weighted strip of options, and you know what contango and backwardation do to a futures curve. Now meet the most-quoted, most-misunderstood number in markets: the VIX. CNBC calls it the “fear gauge,” traders call it a lottery ticket, and almost everyone gets what it is wrong. It is not a forecast of how much the market will move, it is not something you can buy, and the products built on top of it have a nasty habit of grinding to zero or detonating overnight. By the end of this lesson you’ll know exactly what the VIX measures (a variance-swap rate you’ve already learned to build), why its futures almost always slope upward, and why that upward slope is a slow-acting poison for long-vol ETPs — and was a same-day execution for the short-vol ones. Let’s take the gauge apart.

What the VIX actually is

Analogy. Think of the VIX as the price tag on a month of insurance against the S&P 500 swinging around — quoted not in dollars, but in annualized volatility points. When everyone wants insurance, the tag goes up; when nobody does, it sags. Crucially, a price tag is not a prediction: the cost of fire insurance tells you what the market charges to bear fire risk, not whether your house will actually burn. The VIX works the same way.

The definition. The VIX is the 30-day implied volatility of the S&P 500 (SPX), computed model-free from a strip of out-of-the-money SPX option prices. It is not pulled out of Black–Scholes one option at a time; it is built the exact way you priced a variance swap in the prior lesson: take a whole ladder of OTM puts and calls, weight each one by $\frac{1}{K^2}$ (heavier weight on lower strikes), sum them up, and you’ve replicated the fair 30-day variance of the index. The VIX is then just the square root of that variance, annualized and multiplied by 100 to put it in familiar “vol points.”

So the headline you should burn in:

$\text{VIX} = 100 \times \sqrt{\text{30-day variance-swap rate, annualized}}$

The VIX is the square root of a 30-day variance swap rate. That’s it. Every one of those OTM option prices is itself a market-supplied number, so no pricing model is assumed — that’s what “model-free” means. (CBOE actually blends the two nearest monthly/weekly expiries to pin the horizon at a constant 30 days, but the engine is the variance strip you already know.)

Quick worked check of the rule of thumb. Since $\sqrt{252} \approx 15.87$, a VIX of 16 gives a daily move of $16 / 15.87 \approx 1.01\%$. Want a 30-day move instead? Scale the annual 16% by $\sqrt{30/365} \approx 0.287$, which gives about a 4.6% expected move over the next month — nowhere near the “16%” the myth claims.

The fear gauge and its inverse relationship with stocks

The VIX earned its “fear gauge” nickname because of one robust, decades-old pattern: it moves opposite to the S&P 500, and it does so asymmetrically. When the index sells off, the VIX leaps; when the index grinds higher in a calm rally, the VIX drifts sleepily lower.

Why the inverse relationship? Two reinforcing forces:

1. Demand for protection. When stocks tumble, portfolio managers and traders pile into SPX puts as insurance. That surge in demand bids up option prices, and since the VIX is just the price of those options re-expressed as volatility, the VIX shoots up. Calm rallies create no such scramble, so the VIX leaks lower as protection gets cheap.
2. Skew. SPX options carry a permanent volatility skew — out-of-the-money puts trade at higher implied vol than equivalent calls, because crashes are faster and scarier than melt-ups (markets “take the stairs up and the elevator down”). The VIX strip is dominated by these richer downside puts, so anything that fattens the left tail lights up the VIX.

Some rough levels worth memorizing, because the number is meaningless without a reference frame:

The other defining trait: the VIX mean-reverts. It cannot trend to infinity or stay pinned at 80 — fear is exhausting and protection is expensive, so spikes decay back toward the long-run average (historically the mid-to-high teens) over days and weeks. Hold that thought: mean reversion is the single fact that shapes the entire futures curve we’re about to meet.

You can’t buy the VIX — only its futures

Here is the trap that snares every newcomer: the spot VIX is not investable. It’s a calculation — a number CBOE prints every 15 seconds by crunching live SPX option prices. There is no pool of assets behind it, nothing to deliver, nothing to hold in a brokerage account. You cannot buy “the VIX at 15” the way you buy a share at $15.

What you can trade are VIX futures (launched 2004) and options on those futures. And a VIX future is a very different animal from the spot index:

• A VIX future settles to where the spot VIX is expected to be on that future’s expiry date, not to today’s spot.
• Because the VIX mean-reverts, a future expiring in five months is priced near the long-run average VIX, largely ignoring today’s reading.
• So when spot VIX is unusually low, futures sit above spot (the market expects vol to rise back to normal); when spot is unusually high, futures sit below spot (the market expects the storm to pass).

Line those futures up across expiries — one month, two months, three months out — and you get the VIX term structure: a curve of expected future volatility. That curve, not the spot number, is what you actually trade, hedge, and (often) lose money on.

Contango, backwardation, and roll cost

Now the payoff of everything above. Most of the time — historically the large majority of trading days — the VIX futures curve is in CONTANGO: upward-sloping, with near-dated futures below far-dated ones. The reason is exactly the mean reversion we established: spot vol is usually low, and the market prices it drifting back up toward the long-run average, so each later contract is dearer than the one before it.

That upward slope is a slow tax on anyone who stays long VIX futures. To maintain a constant-maturity long position, you must roll: as the front contract nears expiry you sell the cheap expiring future and buy the dearer next one — buy high, sell low, every single month. Worse, each contract you hold converges down toward the lower spot as expiry approaches. Both effects drain value. This is negative carry (a.k.a. negative roll yield), and it bites even if spot VIX never moves a single point.

Flip the market into a panic and the curve inverts into BACKWARDATION: spot VIX spikes far above the futures, so the curve slopes downward. Now the long holder rolls down the curve into cheaper contracts, and each held future converges up toward the elevated spot — positive carry. The cruel catch: you only collect this when fear is already sky-high, which is usually far too late to climb aboard cheaply.

Toggle the regimes below and watch the front future animate toward the spot reference line — and watch the monthly roll P&L flip sign with the curve’s shape.

Worked example — the roll cost in contango. Suppose spot VIX is 16, the front-month future trades at 18, and the next-month future trades at 20 (a textbook upward, contango curve). You’re long one front future, betting on a spike. The spike never comes; vol just sits there. As the front contract marches to expiry, it must converge to spot (16):

Even though spot VIX never budged from 16, you bled 2 points on the first contract’s convergence and you’re now holding a future at 20 that has even further to fall. That’s the negative roll yield in action: in a quiet, contango market a long VIX-futures position is a melting ice cube. (In backwardation the table flips: a front future trading below spot converges up, handing the long holder positive carry.)

Why long-vol ETPs decay (VXX) and the short-vol blow-up (XIV / Feb 2018)

Now connect the roll cost to the products retail traders actually click “buy” on. Exchange-traded products (ETPs — ETNs and ETFs) repackage VIX futures into something that trades like a stock. The headline example is VXX (and its successors), a long-volatility ETN that holds a constant-maturity strip of front VIX futures — roughly the first- and second-month contracts — and rolls a slice of them every day to keep the average maturity pinned near 30 days.

Stack that against the previous section and the tragedy writes itself: because the curve is in contango most of the time, VXX’s daily roll is a daily dose of negative carry. It is structurally buying the dearer back month and selling the cheaper front, every day, while each contract converges down toward a lower spot. The result is relentless decay: over years VXX trends toward zero and has to be reverse-split repeatedly (a 1-for-4 here, a 1-for-5 there) just to keep its share price off the floor. It is not broken — it is doing exactly what it says, and the roll cost is the price of standing perpetually long insurance that usually expires worthless.

So if being long the contango bleeds, the obvious “genius” trade is to be short it — and that’s exactly what the inverse products did.

The short-vol blow-up. Inverse products like XIV (VelocityShares Daily Inverse VIX Short-Term ETN) and SVXY were engineered to return the opposite of the front VIX-futures strip each day. In calm, contango markets they harvested the roll: every day VXX bled, XIV gained. For years this looked like free money — XIV rose roughly tenfold from 2012 to early 2018, and the “short vol” trade became a crowded retail darling.

Then came 5 February 2018 — “Volmageddon.” The S&P fell about 4%, and the VIX more than doubled in a single day (from the mid-teens to ~37), the largest one-day VIX spike on record. Because XIV was levered to the daily move of VIX futures, that doubling translated into a near-total loss: XIV fell about 96% in hours. Its terms contained an “acceleration event” clause that let the issuer terminate it if it lost ~80% in a day — so it was liquidated and shut down days later. A multi-billion-dollar product was gone.

Putting it together

The VIX is the 30-day implied volatility of the S&P 500, built model-free as the square root of a variance-swap rate from a $\frac{1}{K^2}$-weighted strip of OTM SPX options — the exact replication you learned pricing variance swaps. It is annualized vol points, not a tradable forecast: a VIX of 16 implies roughly $\frac{16}{\sqrt{252}} \approx 1\%$ daily moves, not a 16% monthly swing. It is the fear gauge because demand for downside puts and the permanent put-skew make it spike when stocks fall and decay in calm — and it mean-reverts. You can’t buy spot VIX; you trade VIX futures, priced to the expected VIX at their expiry, which line up into the term structure. Mean reversion keeps that curve in contango most of the time, so a long position bleeds negative roll yield every month, while a panic flips it into backwardation and positive carry. That roll cost is precisely why long-vol ETPs like VXX decay toward zero and reverse-split, and why the short-vol harvesters like XIV looked brilliant for years and then died in a day on 5 Feb 2018 — because shorting the contango was always just selling tail insurance.

Next — trading the term structure directly: calendar spreads, the carry-vs-convexity trade-off, and how desks position around the contango that bleeds the longs and the backwardation that burns the shorts.