Chapter 4 · Stochastic Processes in Finance

Quadratic Variation

The odometer of volatility — measuring exactly how jittery a random path truly is.
The big idea

We have seen Brownian Motion as a random walk and GBM as a walk with a trend. But how do we precisely measure how wild and shaky that walk is? Not "it moved a lot" — but an actual mathematical score for the jitteriness? That tool is Quadratic Variation. Think of it as the odometer of volatility: a single number that summarises how much a path has jigged and jagged.

Odometer of volatility gaugeA semicircular gauge labeled QV meter showing the concept of measuring jiggle LOW HIGH QV METER Quadratic Variation = The "Jiggle Score" of any random path Low QV → smooth calm path (σ small)  |  High QV → wild jagged path (σ large)

One number summarises the entire jitteriness of a path.

Step 1 · First-order variation: the ruler that breaks

The natural instinct is to measure the total distance traveled — every up-move and down-move summed together. This is called First-Order Variation. For smooth paths it works. But for a jagged Brownian path, it produces a catastrophic result.

Coastline paradoxThree measurements of a jagged path with decreasing ruler sizes — total length doubles each time approaching infinity Big ruler Medium ruler Tiny ruler Result Total = 80m Total = 160m Total = 320m Explodes to infinity! Like measuring a coastline — the smaller the ruler, the longer the total distance grows

First-Order Variation explodes to infinity on Brownian paths. It is a broken measure.

This is the Coastline Paradox. The coastline of India measured with a 100km ruler appears shorter than when measured with a 1km ruler — the smaller ruler catches every bay and peninsula. Brownian Motion has the exact same problem: measure every micro-wiggle and the total distance becomes infinitely long.

Step 2 · Quadratic variation: squaring saves everything

The fix is elegant: instead of adding raw distances, square each move first — then add them up. Squaring is magical: tiny wiggles nearly vanish (0.01² = 0.0001) while large lunges explode in score (10² = 100). The big violent moves dominate and define the total.

Gentle Drunk vs Violent Drunk — same endpoint, wildly different Quadratic Variation Two walkers shown side by side — the gentle one has small moves squared to small numbers, the violent one has huge moves squared to enormous numbers The Gentle Drunk The Violent Drunk -1010²=100 +88²=64 -1212²=144 -4040²=1600 -7474²=5476 -8686²=7396 QV = 100+64+144+... = LOW — calm, small moves QV = 1600+5476+7396+... = HUGE — wild, violent moves

Both walkers end at the same spot — squaring reveals who was truly wilder along the way.

Here is the most beautiful secret in stochastic mathematics: the Quadratic Variation of a standard Brownian Motion over time t is exactly equal to t. Not approximately — exactly. The path is random, the route unknowable, but the total squared jiggle is perfectly predictable. Order hidden inside chaos.

Step 3 · How we catch σ hiding in a price chart

The GBM formula is dSt = μSt dt + σSt dWt. Volatility σ is invisible on a plain price chart — you only see prices moving. But by computing the Quadratic Variation of returns, you can work backwards and extract σ from real data. Like detecting a car's speed from the skid marks it left behind.

Three-step process to extract sigma from a price chart using Quadratic Variation Left shows price chart, middle shows squaring each return and summing, right shows extracted sigma value Step 1 Observe the price chart Prices going up and down Step 2 Square every daily return Σ (ΔS)² (-2%)² = 0.0004 (+3%)² = 0.0009 (-1%)² = 0.0001 Sum all of them → Step 3 Extract volatility σ σ = √(QV ÷ time) Volatility revealed!

This is how volatility σ is extracted from real market data every single day.

The surface roughness — and the engine warning

Picture the price chart as a 3D landscape. Quadratic Variation is the roughness of that terrain. A calm market looks like a smooth lake surface. A volatile market looks like jagged mountain peaks. The most powerful application: you can detect the terrain becoming rougher before the big crash — like hearing engine vibrations build before the engine blows.

Low QV smooth surface vs High QV jagged surface Left panel shows gentle wavy lines for a smooth low-volatility surface. Right panel shows sharp spiky lines for a high-volatility jagged surface. Low QV — smooth like glass High QV — jagged like sandpaper σ is small — gentle predictable waves σ is large — violent spiky terrain

The roughness of the terrain IS the Quadratic Variation. Watch it rise before prices crash.

A strategist monitoring NSE Nifty in real-time tracks QV as it evolves. When the jiggle-score starts climbing, volatility is building — often before prices make their big move. Like feeling earthquake tremors before the buildings fall.

Try it — the live jiggle-meter

Generate a random price path and watch the Quadratic Variation accumulate in real time. Notice how the QV line grows smoothly and steadily even as the price path zigs and zags — hidden order inside the randomness.

20%
Final QV
Expected (= T)
Realised σ

Green = price path  |  Amber dashed = accumulated Quadratic Variation  |  Purple dotted = expected QV (always = T = 1)

The three measures, side by side
1st
First-Order Variation
Total distance traveled. Explodes to infinity for random paths. Completely useless for Brownian jiggle.
QV
Quadratic Variation
Sum of squared moves. Stays finite. Equals exactly t for Brownian Motion — order inside chaos.
σ
Volatility (Sigma)
The scaling factor of the jiggle. Extracted from QV. The market's true heartbeat, made visible.

In short: First-Order Variation breaks on random paths — it is simply too big. Quadratic Variation squares every move so that big lurches dominate and tiny wiggles vanish. For Brownian Motion, QV always equals time — a perfectly predictable number hidden inside randomness. And volatility σ is the speedometer reading you extract from that QV score. It is how mathematicians listen to the heartbeat of the market.

Next Chapter: Scaled Random Walk →