A contraption from 1876
Sir Francis Galton built a board of pegs. A ball drops in at the top; at every peg it bounces left or right on a pure 50/50 chance. Twelve pegs, twelve coin flips, and nobody on Earth can tell you where one ball will land.
Try it. Drop a single ball and watch its path. Then drop two thousand and watch something strange happen to the pile.
The paradox to sit with: each ball is perfectly unpredictable, and the pile is almost perfectly predictable. Randomness in the small, order in the large.
Now relabel the parts
Here is the entire trick of this page: a month in the stock market is a ball falling through 21 pegs — one peg per trading day. Each day, news nobody saw coming knocks the price up or down a little. Where the ball lands is your month’s total return.
Start with the coin-flip market, where each day is exactly ±1.1% (the S&P 500’s real typical daily move). Then switch to the real thing: every ball you drop relives an actual stretch of history, day by day, from the last ten years. One ball might live through a sleepy August; another gets February 2020.
Three things to notice while you play. First, the real market builds nearly the same bell as the coin-flip machine — that is the whole lesson. Second, the bell’s center sits a touch right of zero: the market’s coin is very slightly loaded upward (about +0.06% a day here — invisible daily, decisive over decades; it’s why patient investing is not gambling). Third, switch from a week to a month and the pile widens — but only by about the square root of the extra time, not in proportion to it. Risk grows slower than time. That little fact runs half of finance.
And try Tesla: same bell shape, roughly three times wider than Coca-Cola’s. Volatility (σ) doesn’t change the machine — it just spreads the pegs.
Think you can call tomorrow?
Below are 60 consecutive real trading days of the S&P 500 from some mystery stretch of the last decade — ripped from the calendar so you can’t cheat. Study the trend, the momentum, the vibes. Then call the next day: up or down? Ten rounds. The coin is watching.
Close to a bell — wilder at the edges
Everything above is true, and now the honest footnote. Here is every single trading day of the S&P 500 from the last ten years — all of them — piled into a histogram, with the best-fitting bell curve drawn on top. The middle fits like a glove. Now flip on the log scale and look at the edges.
Real markets have fat tails: extreme days arrive far more often than the perfect bell predicts — because on the worst days the bounces stop being independent. Panic is contagious; everyone’s peg tilts the same way at once. The most famous example predates our data: on October 19, 1987, the market fell about 20% in one day — an event the textbook bell curve calls essentially impossible. It happened anyway. So: the bell curve is a superb map of ordinary weeks and a poor map of earthquakes — and knowing which map you’re holding is what risk management is about.
What to remember from this page
- Short-term returns are close to unpredictable. Yesterday’s pattern told you almost nothing; a coin flip kept up with you. If tomorrow’s rise were obvious today, buyers would push the price up today — leaving tomorrow a coin flip again. (You’ll meet this idea formally as market efficiency.)
- Unpredictable days build a predictable shape. Many small independent surprises, compounded, produce a distribution close to the bell curve — with a slight upward tilt that rewards patience. You can’t forecast a day; you can characterize a decade.
- “Close to” is doing real work. The real bell is a little lopsided and noticeably fat in the tails. The center is where averages live; the tails are where crises live. Respect both.
Data & methodology
Returns are close-to-close daily percentage changes, stored to the nearest 0.01%. Coca-Cola and Tesla prices are adjusted for splits and dividends (so their returns approximate total returns); the S&P 500 is a price index, so its returns exclude dividends. In Act II, every “real market” ball replays a randomly chosen stretch of consecutive actual trading days; the resulting pile is the distribution of overlapping (rolling) weekly or monthly returns. The coin-flip market moves exactly ±one daily standard deviation of the selected series with probability ½ each day. The fitted curves are normal distributions matched to each pile’s own mean and standard deviation. This page is a teaching instrument, not investment advice.