Tail Events

Tail events are the rare, far-from-average outcomes that — across the sources — dominate results in finance and innovation, even though they are exactly what extrapolation cannot predict. In The Psychology of Money, Morgan Housel makes this the core of his “long tails” chapter: “Long tails—the farthest ends of a distribution of outcomes—have tremendous influence in finance, where a small number of events can account for the majority of outcomes.” His examples: Apple driving ~7% of the index’s returns in 2018; people on “tail projects that drive tail returns” having “tail careers” (Google’s 0.2% acceptance rate). He also stresses how easy they are to underappreciate “because in most fields we only see the finished product, not the losses incurred that led to the tail-success product” — the survivorship-bias link to Post-Mortems.

The concept is the hinge of the wiki’s Forecasting page: most financiers “extrapolate trends from the past, disregarding the risk of extreme ‘tail’ events. Venture capitalists look for radical departures from the past. Tail events are all they care about.” Because the disruptions that create wealth “cannot be predicted based on extrapolations of past data,” tail events are precisely the thing conventional, calibrated forecasting is blind to — which is why the Power Law (not central tendencies) governs venture outcomes. The non-forecasting alternative is Benjamin Graham’s Margin of Safety, whose purpose Housel quotes as “to render the forecast unnecessary.”

Context: In statistics, “tails” are the extreme ends of a probability distribution; “tail events” (overlapping with Nassim Taleb’s “black swans”) are low-probability, high-impact outcomes whose influence is wildly disproportionate to their frequency.

Where this appears

• The Psychology of Money — the “long tails” chapter: a small number of events account for the majority of outcomes (the index, VC, “tail careers”); easy to underappreciate because we only see finished products.
• Forecasting — tail events as the radical departures from past data that extrapolation cannot see and that venture capitalists exclusively target.