Implied Volatility & the Smile

Black–Scholes eats six inputs and spits out a price. You’ve met all six. Here’s the trick nobody tells you up front: in the real world, traders almost never run the formula the way we taught it. They run it backwards — and the number that falls out, implied volatility, becomes the language the entire options market speaks. By the end of this lesson, “price” will feel like an awkward translation, and “vol” will feel like the mother tongue.

Let’s start with a guess.

Five inputs you can see, one you can’t

Line up the six Black–Scholes inputs and sort them by honesty. Spot price? On the screen. Strike? Written in the contract. Time to expiry? Count the days. Risk-free rate? A Treasury yield. Dividends? Announced or forecast with decent confidence. Five of the six are observable facts you can pull up in seconds.

Then there’s volatility, $\sigma$ — the standard deviation of the underlying’s future returns. It refers to how much the stock will wobble between now and expiration, which hasn’t happened yet. You cannot look it up, because the future hasn’t been printed. It’s the one genuinely unknown ingredient in the recipe.

So Black–Scholes, run forwards, has a problem: to get a price you must assume a $\sigma$ you can’t actually know. Awkward — until you notice there’s a seventh number sitting right there on the screen that we’ve been ignoring.

Run the formula backwards

Here’s the inversion, the single most important idea in this lesson. Forwards, Black–Scholes is:

$\text{price} = \text{BS}(\,\underbrace{S, K, T, r, q}_{\text{all observable}},\ \sigma\,)$

Five inputs are known and the price is also known. So there’s exactly one unknown left in the equation: $\sigma$. Instead of feeding in a volatility to get a price, we feed in the price and solve for the volatility that makes the formula spit that price back out.

That solved-for value is the implied volatility (IV): the volatility implied by the market price. It’s the market’s collective forecast of future volatility, reverse-engineered out of what people are actually paying. If a call is trading rich, IV is high — the crowd is bracing for a big move. If it’s trading cheap, IV is low — the crowd expects calm.

There’s no neat formula — so we hunt for it

You’d love a clean algebraic inverse, $\sigma = \text{something}$. It doesn’t exist. Black–Scholes can’t be rearranged to isolate $\sigma$ in closed form — the volatility is buried inside two cumulative-normal terms. So we solve it numerically: guess, check the resulting price, adjust, repeat until it matches.

Two facts make this easy and reliable:

• Price is strictly increasing in $\sigma$. That’s just vega > 0 — every option you’ve met gets more expensive as volatility rises. Strictly monotonic means there’s exactly one IV for any sane price, and no ambiguity about which way to nudge your guess.
• Vega is the slope. It tells you how fast price climbs per point of vol, which turns guessing into a precise step.

The crude-but-bulletproof method is bisection: bracket IV between, say, 1% and 200%, try the midpoint, and halve the interval based on whether your trial price is too high or too low. Slow but it always converges.

The fast method is Newton’s method, which uses vega as the slope to jump straight toward the answer:

$\sigma_{n+1} = \sigma_n - \frac{\text{BS}(\sigma_n) - \text{market price}}{\text{vega}(\sigma_n)}$

In English: take your current guess, see how far the model price misses the market price, and divide that miss by vega to convert the price error into a volatility correction. Then step.

Implied vs realized: two very different volatilities

People say “volatility” and mean two opposite things, so pin them apart now.

Realized (historical) volatility is how much the stock actually wobbled. You measure it after the fact from the price record — annualize the standard deviation of past daily returns. It’s a fact about the past, fully observable, no opinion involved.

Implied volatility is the market’s forward-looking bet, the number we just extracted from option prices. It’s about the future, so it’s an expectation, not a measurement.

These two numbers are usually not equal — and the gap has a direction. On average, across most markets and most of the time, implied vol sits above subsequently realized vol. Option sellers, who eat catastrophic losses when a calm stock suddenly gaps, demand to be paid extra for bearing that tail risk. That premium baked into option prices is the variance risk premium — the structural reason IV tends to run “rich” relative to what actually unfolds.

The smile: Black–Scholes’ big confession

Now the punchline the whole lesson has been building toward. Black–Scholes assumes a single constant $\sigma$ that applies to every strike on a given underlying and expiry. If that assumption were true, here’s the prediction: back out IV from every strike — deep out-of-the-money puts, at-the-money, far out calls — and you’d get the same number every time. A flat, boring horizontal line.

The market does no such thing. Take real option prices, invert each one, and plot implied vol against strike (or moneyness, $K/S$). You get a curve. Different strikes trade on different implied vols. This is the volatility smile, and it is the market loudly confessing that the one-constant-$\sigma$ model is wrong.

Play with the shape below. The dashed flat line is the Black–Scholes fantasy; the curve is what markets actually quote. Toggle the presets to see the two canonical personalities.

Two shapes show up again and again:

• Equity-index skew (the smirk). For stock indices, IV is highest for low strikes — the out-of-the-money puts — and slopes downward as strike rises. The curve leans left. Two forces drive it: relentless demand for crash protection (institutions perpetually buy downside puts, bidding their vol up), and the fact that real index returns have a fatter left tail than the lognormal model allows — crashes are sharper and more frequent than a bell curve predicts. This skew barely existed before October 1987; after Black Monday’s one-day collapse, it appeared sharply and never left. Traders literally call it the “1987 smirk.”
• FX smile. For currency pairs, the curve is roughly symmetric — both wings lift up. The reason is structural: in a currency pair, either side can crash. A big move up in EUR/USD is a big crash in USD/EUR, so the market demands tail protection on both ends. Symmetric fear, symmetric smile.

Remember the lognormal picture from the pricing core? That bell-ish, right-skewed distribution is exactly what Black–Scholes assumes the terminal price obeys. The smile is the market telling you the true distribution has heavier tails than that — especially on the downside for equities. Watch the lognormal shape, then imagine fattening its left tail: those fatter tails are precisely what make out-of-the-money options worth more than the flat model says, which is what shows up as elevated IV at those strikes.

Term structure: vol changes with time, too

The smile shows IV varying across strike. But hold strike fixed and walk out along expiration instead, and IV changes again — that’s the term structure of volatility. A one-week option and a one-year option on the same name routinely imply different vols. In calm markets the term structure usually slopes up (longer horizons price in more uncertainty); right after a shock it can invert, with near-dated options screaming high vol that’s expected to fade.

Put the two dimensions together — IV as a function of both strike and expiry — and you get the volatility surface: a 3-D sheet that traders fit, smooth, and re-mark all day. It’s the real object banks manage; the single-number “the vol is 20%” is a convenient fiction.

The VIX: fear, in one number

The most famous IV figure on Earth is the VIX. It’s a model-free measure of the 30-day implied volatility of the S&P 500, built not from one option but from a whole strip of S&P 500 option prices blended together — so it doesn’t depend on Black–Scholes being right. Quoted as an annualized percentage, it’s nicknamed the “fear gauge”: it sits in the low-to-mid teens in placid markets and spikes violently in crashes, when everyone stampedes to buy protective puts and option prices — hence implied vol — explode. When you hear “the VIX hit 80,” that’s the options market pricing in chaos.

Speaking vol: how traders actually think

Here’s the payoff for all this machinery: professionals quote, hedge, and reason in volatility, not price. A dealer doesn’t say “I’ll sell that call for $5.00”; they say “I’ll sell it at 25 vol.” Price is a derived afterthought — feed your IV into Black–Scholes and it pops out.

That reframes everything you do with options:

• Buying any option is going long volatility. You profit if realized vol — or the IV others will pay you later — comes in higher than the IV you paid. Selling is the reverse: short vol, hoping for calm.
• “Cheap” and “expensive” mean low and high IV relative to your own forecast, not low or high dollar price. A $0.20 option can be wildly expensive (40 vol when you think 20 is fair); a $15 option can be a bargain.
• The smile means even “the vol” is a question of which strike. Quoting a number without a strike and expiry is like quoting a price without a currency.

Putting it together

You can now run Black–Scholes in the direction that matters. Five inputs are observable, so fixing the market price lets you invert for the one that isn’t — implied volatility, found numerically because vega keeps price strictly increasing in $\sigma$. IV is a forward-looking bet that usually runs above realized vol (the variance risk premium). And because the model assumes one constant $\sigma$ but markets quote a different IV at every strike, you get the smile — left-leaning skew for equities, symmetric for FX — extended across expiries into the full volatility surface, with the VIX as its most famous single reading.

That closes the loop on pricing: you can now read Black–Scholes both directions, and you speak the market’s native language of vol, skew and surfaces. Everything in this arc — the binomial intuition, the formula, the Greeks, and now implied volatility — comes together next in the topic’s final exam, a single graded run to prove it stuck.