A thought experiment from the savannah: your whole group suddenly sprints left. You can stand still and evaluate the evidence for lions, or you can run first and evaluate later. For a hundred thousand years, the runners out-reproduced the evaluators. You are descended from the runners.
Herding, copying the group under uncertainty, is not stupidity. Under genuine ignorance it is often the best available algorithm: the crowd aggregates information you don’t have. That is exactly what makes it treacherous in markets, where the crowd’s behavior changes the price, and the price changes what the crowd believes.
The loop that builds bubbles
Watch the machine run. An asset rises. Early buyers get rich, visibly (riches are always visible; the neighbor’s new car parks outside). Observers infer information from the price itself: so many people can’t be wrong. They buy, which lifts the price, which strengthens the inference for the next observer. Economists call it an information cascade: each person rationally weighs the crowd over their own doubts, and private information stops entering the price entirely.
The cruel geometry is that the signal feels strongest at the worst moment. Maximum consensus arrives when everyone who could buy has bought — which is, mechanically, the top. The 1999 tech bubble, the 2007 housing chain letter, the crypto manias: different assets, same choreography, including the final step where the last arrivals, who felt safest because the crowd was largest, absorb the losses.
The mirror image runs at bottoms. In a crash, the crowd sprints for the exit, and standing still feels reckless even to owners of boring index funds who wrote plans for precisely this day. Selling in 2009 felt prudent. That is herding wearing a suit.
The modern accelerant
Your feed made the ancient loop faster. Social media compresses the neighbor’s-new-car effect from years to hours, and its algorithms curate winners: you see the screenshots of gains, never the quiet liquidations. FOMO, the fear of missing out, is simply herding’s emotional interface, and it now updates in real time. When an asset’s main argument is that everyone is talking about it, note that this argument, historically, has the worst timing of any argument in finance.
Standing still on purpose
You cannot out-argue a sprinting crowd in the moment. You can pre-decide your relationship to it.
Let the plan absorb the pressure. The standing monthly buy and the written policy are herd-proofing: they buy through panics and refuse manias without requiring courage from you on the day. The rebalancing rule goes further — it mechanically leans against the herd, trimming whatever the crowd just inflated.
Reclassify buzz as data. When a tip arrives socially (group chat, dinner party, feed), practice one reframe: popularity is not evidence about the asset; it is evidence about where in the cycle you are hearing about it. By the time an opportunity reaches you through enthusiasm, you are late by construction.
Budget the itch. If you know you will sometimes need to run with a crowd, give that impulse a fixed, small sandbox — money whose total loss changes nothing — and keep it walled off from the plan. A contained indulgence beats a chronic leak.
The crowd is not always wrong. It is wrong at exactly the moments when agreeing with it feels most comfortable — and investing pays whoever can be comfortable being uncomfortable.