Fat Tails & Black Swans: When the Rare Runs the Show

You already learned to think in probabilities — to trade the yes/no switch for a dial, anchor on base rates, weigh bets by expected value, and respect ruin. This course is where that dial meets the part of reality it was built for: the wild domains, where the rules of the mild ones quietly stop applying.

Here is the whole problem in one sentence. In some worlds, no single event can move the total much — your height, the temperature tomorrow, the weight of the next person through the door. Add a thousand of them and the average is rock-solid; the biggest one is barely larger than the typical one. Call that world Mediocristan. In other worlds — wealth, book sales, war deaths, market moves, pandemics, city sizes — a single observation can be larger than all the rest put together. One bestseller outsells ten thousand titles combined; one bad day erases a decade of gains; one outlier is the dataset. Call that world Extremistan. Almost all of our intuitions, our averages, and our tidy bell curves were built in Mediocristan — and they get people killed in Extremistan.

A fat tail is the technical name for that second world: a distribution where extreme values are far more frequent, and far more extreme, than a normal bell-curve intuition expects. The course builds the idea from the ground up: the two worlds and why one sample can or can’t dominate the total; the normal distribution and exactly how it under-counts the extremes; power laws and the 80/20 family that generate fat tails; why the mean and standard deviation mislead when the tail is fat (and why backtests, averages, and “it’s never happened” all lie); black swans — rare, high-impact, and rationalized only after the fact — and the turkey problem that breeds them; and finally ergodicity and ruin: why, in a fat-tailed world, the first rule is to never be exposed to the one outcome you can’t come back from. You’ll drive an interactive sampler that lets you fatten the tail with a slider and watch a single draw swallow the whole total. By the end you’ll have the rarest skill in forecasting — knowing which world you’re in before you trust the average — and the doorway to calibration, the expert capstone of the probability path.