What broke on 15 January 2015?

On one morning in January 2015 a major developed-market currency moved by more in a few minutes than it normally moves in years, several regulated brokers were rendered insolvent, and thousands of clients discovered they owed money they did not have. It is the best available factual illustration of everything in this unit, and it happened without a single default, war, or scandal.

The setup

On 6 September 2011, with the franc surging as the euro-area crisis deepened, the Swiss National Bank announced a minimum exchange rate of CHF 1.20 per euro. Its language was deliberately absolute: it would enforce the floor "with the utmost determination" and was prepared to buy foreign currency "in unlimited quantities."

For over three years this worked. EUR/CHF sat just above 1.20 and traded in a very narrow range. Implied volatility on the pair fell to multi-year lows, because the market had come to treat the floor as a physical constraint rather than a policy choice. Leveraged positions accumulated on the assumption that the downside was capped at a known level.

The cost was on the SNB's balance sheet. Defending a strong-side floor means creating francs and buying euros, and by the end of 2014 the SNB held roughly CHF 495 billion in foreign-currency investments — on the order of three-quarters of Swiss GDP. Meanwhile the ECB was visibly moving toward large-scale asset purchases, which it announced on 22 January 2015. Holding the floor through that would have required buying euros in genuinely unlimited size.

The morning

At about 09:30 CET on 15 January 2015, the SNB announced it was discontinuing the minimum exchange rate. Simultaneously it cut the interest rate on sight deposits by 50 basis points to −0.75%, moving its target range for three-month Libor to between −1.25% and −0.25%.

What followed was not a fall but a vacuum. With the floor gone there was no natural bid, and EUR/CHF traded down through 1.00 to prints around 0.85 and below on some venues in extremely thin liquidity, before recovering to close the day near 1.04. The franc ended the session roughly 15–20% stronger against the euro. The Swiss Market Index fell 8.7% on the day, its largest single-day decline since 1989.

Why stop-losses did not help

This is the operational lesson, and it is worth stating precisely. A stop-loss is an instruction to trade when a level is reached. In a gap, there are no prices at that level — the market simply reprints far below it. Orders placed at 1.19 were filled in the 0.90s.

For leveraged retail clients that meant losses exceeding their deposited capital, and for brokers it meant client accounts with negative balances that the broker was legally obliged to cover.

• FXCM reported roughly $225 million of negative client equity and required a $300 million rescue loan the following day.
• Alpari UK entered insolvency.
• Global Brokers NZ closed.
• Several large banks reported FX losses in the tens to low hundreds of millions of dollars.
• The SNB itself reported an annual loss of CHF 23.3 billion for 2015.

The four things it teaches

1. "Unlimited" describes willingness, not ability — and willingness is precisely the variable that changes. The SNB never ran out of francs. It changed its mind about the cost.
2. A defended level suppresses measured risk while accumulating actual risk. Low realised volatility was a consequence of the intervention, not evidence that the pair was safe.
3. Liquidity is a state, not a property. The most liquid market on earth had no bid for several minutes.
4. Leverage converts a price move into a solvency event — for clients, and then for the firms that faced them.

This is a factual account of a documented past episode. It is not a forecast, not a claim about any current regime, and not advice about anything.

In the data

A day like that reaches a daily price history as a single row, with the entire move sitting in one number. The three days around it are below.

The 15 January 2015 row opens at 1.2008, carries a low of 0.8643 and records its close at 1.2008, the floor, although the pair ended that day near 1.04; the collapse only reaches the closing prices the next morning, at 0.994. A close-to-close series therefore shows nothing at all happening on the day it happened. Nor does anything in the row say whether that low was a level anyone was filled at or a print that existed for a moment. Daily bars are the wrong resolution for a discontinuity, and they give no sign that they are.

Try it now

1. EUR/CHF over the longest window this page holds is below, as candles. Find 15 January 2015 — it is not hard to find.

2. Measure two things and put the numbers next to each other: the width of the trading range between September 2011 and January 2015, and the single-day move on 15 January 2015 from its open to its low. Drop a Level at 1.20 to see what the first number was actually made of.
3. The floor was held with an open-ended promise to print francs, and abandoned alongside a policy rate cut to −0.75%. Write one neutral sentence on what the pair's calm looked like from the inside — and what that calm was actually measuring.