Chapter 5 · Stochastic Processes in Finance

Scaled Random Walks

The bridge chapter — how a simple coin flip, flipped fast enough, becomes Brownian Motion.
The bridge

Everything we have studied so far — Brownian Motion, GBM, Quadratic Variation — rests on a single foundation. This chapter builds that foundation. It shows how the simplest game imaginable, flipping a fair coin, transforms into the smooth, continuous mathematics of the Bell Curve when you flip it fast enough and small enough. This is the bridge between basic probability and quantitative finance.

Bridge diagram from coin flip to Brownian Motion Three boxes connected by arrows — coin flip on the left, scaled random walk in the middle, Brownian Motion on the right Coin Flip ±$1 per minute 50/50 odds Scaled Walk Faster + smaller steps More and more flips Brownian Motion Continuous, smooth Normal distribution flip faster steps → 0

Three stages — and each one is just the previous stage run faster and smaller.

1 · The symmetric random walk: the coin flip game

Imagine you are standing at a stock price of $100. Every minute, you flip a fair coin. Heads means the price moves up $1. Tails means it moves down $1. This is the Symmetric Random Walk — symmetric because the odds are perfectly 50/50 in each direction. Nothing more, nothing less.

Symmetric Random Walk — staircase path from coin flips A staircase-style price chart showing the result of ten coin flips: H T H H T H T T H T, ending at $102 $104 $103 $102 $101 $100 H H T H H T T H T H +$1 +$1 -$1 +$1 +$1 -$1 -$1 +$1 -$1 +$1 $102 Each minute = one coin flip. Each flip = ±$1.

The staircase shape is the signature of a discrete random walk — sharp corners, fixed step size.

The Independence Property

Each step is completely independent. Whether you got Heads last time has zero effect on the next flip. The coin has no memory. Neither does the market.

The Martingale Property

If the price is $105 right now, your best guess for where it will be in an hour is still $105. Not higher, not lower. The game is perfectly fair — no free lunch, no edge.

The Martingale Property is one of the most important ideas in quantitative finance. In an efficient market, the current price already contains all available information. There is no pattern to exploit without taking on additional risk. Today's price is the best prediction of tomorrow's price.

2 · The scaled random walk: turning up the resolution

The stock market does not move once per minute. It moves thousands of times per second. Scaling means two things happening simultaneously: flipping the coin faster (more steps per second), and making the step size smaller (moving fractions of a penny instead of whole dollars). As you do both, something remarkable happens to the shape of the path.

Three resolutions of a random walk — coarse staircase, medium jagged, fine smooth Three panels showing the same random walk at increasing resolution — 10 steps shows a staircase, 100 steps shows zigzag, 1000 steps approaches a smooth curve 10 steps Coarse — staircase 100 steps Finer — jagged line 1,000+ steps Smooth — curves! Δt = 6 min, Δx = $1 Δt = 36 sec, Δx = $0.10 Δt → 0, Δx → 0

Same process, three zoom levels. As steps shrink to zero, the staircase melts into a smooth curve.

In your Price Landscape, tick data is the raw staircase — the coarse random walk. 1-hour candles are already in the scaled-limit regime — the jagged edges have averaged out into something approaching the smooth Bell Curve. You are literally choosing your resolution every time you pick a timeframe.

3 · The magic moment: where the bell curve is born

Here is the part that feels like magic. As you make the steps faster and smaller, the final position of your random walker starts following the Normal Distribution — the Bell Curve. Every individual step was just a boring coin flip: Up or Down. But the sum of thousands of those coin flips produces the most famous shape in all of mathematics. This is the Central Limit Theorem in action.

Central Limit Theorem in action — few flips gives flat distribution, many flips gives Bell Curve Two panels: left shows a flat histogram from 10 flips, right shows a bell-shaped histogram from 1000 flips 10 coin flips Final position — rough, uneven 1,000 coin flips Final position — the Bell Curve! Final position Jagged — no clear shape yet Final position The Bell Curve emerges!

Add enough independent random steps and the Bell Curve always appears — this is the Central Limit Theorem.

The Central Limit Theorem says: add together a large number of independent random variables — it doesn't matter what shape each one has — and their sum will always approach the Normal Distribution. Each coin flip was binary: only Up or Down. But add thousands together and the Bell Curve is inevitable.

4 · The log-normal distribution: prices can never go below zero

There is one final problem to fix. A plain random walk can go negative — the price could theoretically hit –$50. But a stock price can never be negative. You cannot owe someone a company. GBM solves this by applying randomness to the returns (percentage changes), not the raw price level. And when you apply the Bell Curve to returns, the resulting price follows a different distribution — the Log-Normal Distribution.

Normal distribution for returns vs Log-Normal distribution for prices Left panel shows a symmetric bell curve centered at zero for returns. Right panel shows a right-skewed log-normal curve starting at zero for prices. Stock Returns — Normal Stock Prices — Log-Normal 0% -returns +returns Symmetric — can be positive or negative $0 Hard floor! Price → Skewed right — price can never go below $0

Returns are symmetric around zero. But prices, being the exponent of those returns, are bounded at zero.

The elegant logic: if a stock's daily return follows a Normal Distribution (sometimes +2%, sometimes -2%), then the stock price itself — which is what you get by multiplying those returns together over time — follows a Log-Normal Distribution. It can grow to infinity upward, but it can never cross zero downward. Exactly like a real company's share price.

Try it — watch the bell curve being born

Run many random walks simultaneously and watch where they end up. With just a few walks, the final positions look random and scattered. Run thousands and the Bell Curve emerges from nowhere. Select the resolution to see how scaling changes the path shape.

500
Simulations run
Mean final pos.
Std deviation
Bell curve fit?

Left: sample paths  |  Right: histogram of final positions — watch it become a Bell Curve!

The four key ideas
±1
Symmetric Random Walk
Coin flip game. Each step ±$1. Independent. Fair — no pattern, no edge. The Martingale Property.
↑↑↑
Scaled Random Walk
Faster flips + smaller steps. As resolution increases, the staircase melts into a smooth path.
CLT
Central Limit Theorem
Add enough independent random steps and the Bell Curve is inevitable — regardless of the shape of individual steps.
ln
Log-Normal Prices
Apply the Bell Curve to returns (not prices) and the resulting price distribution is Log-Normal — it can never go below zero.

In short: Flip a fair coin. If you flip it fast enough, and make each step small enough, the jagged staircase transforms into the smooth, continuous mathematics of the Bell Curve. This is the Central Limit Theorem — and it is the proof that Brownian Motion is a valid and mathematically rigorous model for how markets move. The bridge between a simple coin flip and a hedge fund's pricing model is just a matter of resolution.

Next Chapter: The Markov Property →