73 corrections or 339? The counting convention decides
Anchor drawdowns at all-time highs and a 98-year history has 73 episodes of 5%+; anchor them at local peaks and it has 339. Neither is wrong — they answer different questions. Never quote a drawdown count without its convention.
One dataset, two answers
Count S&P 500 declines of 5% or more from 1927 to 2026 and you can defensibly report 73 — or 339. Anchored at all-time highs, a new episode begins only after the prior peak is fully recovered, so a decade like the 1930s is one long episode. Under the practitioner convention, every local peak that gives back 5% starts a new event, so the same decade splinters into dozens.
Why it matters
The two conventions answer different questions. ATH anchoring measures time under water relative to your high-water mark — the number that matters for an investor’s account statement and for recovery-time statistics. Local-peak counting measures how often the market interrupts an advance — the number that matters for trade management and hedging cadence (about 3.4 events of 5%+ per year on this sample). Confuse them and you either overstate calm or overstate chaos.
The discipline
State the convention next to every drawdown count, and pick it by the question: account risk → ATH anchoring; tactical interruption frequency → local peaks. When you read “the market corrects every N months”, the first thing to check is not the data source — it is the counting rule.