When The Exchange Or Your Line Goes Down

There are two very different problems that both look like "I can't trade": the exchange itself has gone down, and your own connection to a still-functioning exchange has broken. Confusing the two wastes the first, most valuable minutes of an outage, because the right response is almost opposite in each case — an exchange-wide halt means the whole market is frozen and there's nothing to reconnect to yet, while a connectivity loss means the market is moving without you and every second of delay is a second your positions are unhedged against a market you can't see.

The first move in either case is the same: figure out which one it is, fast, using a source that doesn't depend on the broken connection — a phone call to the broker, a different market data feed, a colleague on a different network. Once that's established, a connectivity loss becomes an operational fire drill with a known playbook: switch to a backup connection or a voice broker to manage existing risk, confirm which working orders are still live at the exchange versus canceled by the disconnect (this varies by venue and matters enormously — some exchanges cancel all resting orders on a dropped session, others leave them live), and avoid sending new orders through an unreliable path where you can't be sure whether a cancel or a new order actually reached the exchange.

Worked example

A desk's primary market data and order feed drops mid-session while holding a delta-hedged options book. The position isn't the immediate danger — the danger is that the hedge, which needs continuous rebalancing against the underlying's price, is now blind. Within two minutes the desk confirms via a backup terminal that the exchange itself is fine and the problem is local connectivity, switches order flow to a secondary broker connection kept live for exactly this purpose, and re-establishes a market data feed to see current prices before sending any new orders. Only after confirming what's actually live at the exchange — by requesting a position and open-order reconciliation from the broker — does the desk resume active hedging, rather than guessing at what orders might still be sitting in the book and potentially doubling up on a hedge that was never canceled.

The mistake that turns an outage into a bigger loss is sending orders through a connection you're not sure is reliable, "just to be safe" — an order that seems to fail silently might have actually gone through, and now there are two positions where the desk thinks there's one.