Market-Wide Circuit Breakers

Individual stocks have mechanisms to pause trading when their own price moves too fast (limit up-limit down bands). But a genuinely broad, market-wide panic — a selloff hitting essentially every stock at once — calls for something bigger than pausing names one at a time. Market-wide circuit breakers halt trading across the entire US equity market simultaneously, triggered off a single benchmark: the intraday percentage decline of the S&P 500 from its previous close.

There are three tiers. A Level 1 decline of 7% triggers a 15-minute market-wide halt. A Level 2 decline of 13% triggers another 15-minute halt. A Level 3 decline of 20% halts trading for the remainder of the day — there's no reopening once that threshold is hit. Level 1 and Level 2 halts only apply if the decline occurs before 3:25pm Eastern; after that time, the market is close enough to its scheduled close that a further pause isn't triggered for those tiers, since there's little session left for it to protect. The Level 3, 20% threshold applies at any time during the trading day.

The logic is that a decline this fast and this broad is more likely to reflect a temporary imbalance — panic selling, a liquidity air pocket, algorithms all reacting to the same signal at once — than a instant, permanent repricing of the entire economy's worth. A short, coordinated pause gives market participants a moment to assess what's actually happening, for stale or erroneous information to be corrected, and for buyers who were caught flat-footed to reassess and step back in, before trading resumes.

Worked example. The S&P 500 opens the day and, by 10:15am, has fallen 7.2% from the prior close. This trips the Level 1 threshold: every US equity exchange halts trading simultaneously for 15 minutes. Trading resumes at 10:30am. If the index falls a further 6% from there, cumulatively reaching a 13% decline from the prior close, the Level 2 halt triggers, pausing the market for another 15 minutes — occurring in the same session because each tier is measured against the same fixed prior-close reference, not against the level right before the previous halt.