The midnight spread spike — why your stop gets hit at rollover

Marek closed his platform at 10:50pm with an open short position on EUR/USD and a stop loss eight pips above price. In the morning he saw the position had been stopped out at 11:01pm — even though the candle from that minute never touched his level, not even by half a pip. His first thought was that the broker hunts stops. The truth is less dramatic and far more useful. Marek had run into something traders have nicknamed the midnight spread spike: a brief window at the turn of the day when liquidity vanishes and the spread widens many times over. Below I explain where it comes from and how to set stops that walk through the window unharmed.

What the midnight spike is and why it lands near the day's end

The midnight spike is a trader-forum name for the window roughly between the close of the New York session and the lull before Asia opens. The reference point is 5pm New York time — the moment when most brokers close the trading day and recalculate swap, the points charged for holding a position overnight. Several things land at once at that minute. The United States session is fading, the largest institutional players are squaring their books, and a good stretch of time still separates the market from the liquid Asian session. For anywhere from a dozen to several dozen minutes the market turns thinner than at any other point in the day.

One calendar caveat matters: the exact time drifts. Europe and the United States change their clocks on different dates, so for a few weeks each year the window lands an hour earlier or later than usual. That is why the spike is an approximate label — the real moment has to be checked at your own broker, because it is the broker that defines the rollover time on its server.

Where the spread widening comes from

The spread is the gap between the best buy price (the bid) and the best sell price (the ask) that a broker receives from its liquidity providers — banks and market makers. Those prices do not come from nowhere; someone on the other side has to post an order. When institutions pull their quotes at the close of the trading day, the order book empties out and the gap between the nearest buy and the nearest sell automatically grows. On liquid EUR/USD the spread can sit at a fraction of a pip during the day and briefly jump to a dozen pips in the rollover window. On exotic or cross pairs the gap can be far more brutal.

The effect is well documented. In its studies of foreign exchange market microstructure, the Bank for International Settlements shows plainly that the spread widens when trading volume and liquidity fall — and both are at their lowest precisely at the seam between sessions. In other words, this is not an anomaly or a conspiracy but a predictable consequence of how the market breathes across the day. If you want to understand how liquidity is distributed between sessions, a good place to start is the article on the New York session and its close, and the broader picture of intraday liquidity sits in the trading hours section on forexmechanics.com.

"The best time to trade is during the hours when the market has the most liquidity. Entering a position when volume dries up means worse execution prices and wider spreads." — Kathy Lien, Day Trading and Swing Trading the Currency Market, Wiley, 2016.

Why a tight stop loss gets hit when price never reached it

Here is the part that frustrates traders the most. The chart on your platform usually plots the bid price. A stop loss on a short position, however, is in effect a buy order, so it executes at the ask. Under normal conditions the bid and the ask sit a fraction of a pip apart, so you never even notice the difference. But when the spread widens to a dozen pips, the ask pulls clearly away from the bid, upward.

And that is exactly when the seemingly impossible happens: the bid printed on the candle stops just below your stop level, while the ask spikes a few pips through it, and it is the ask that triggers the order. In the morning you look at a historical bid chart that never touched the line and you have every right to feel cheated. In reality it was the ordinary mechanics of two prices in a thin market. For a long position it is the mirror image — the bid falls and triggers the stop while the ask on the chart looks innocent. If the bid-ask split is new to you, a useful companion is the piece on price slippage, because it is the same phenomenon seen from the angle of order execution.

An illustrative example — a scalp on EUR/USD at the rollover

Let us walk through Marek's situation step by step. To be clear: this is an illustrative example showing the mechanics, not a record of a real trade.

Marek goes short on EUR/USD at 1.0850 around 10:40pm, expecting a quick move down. He sets a stop loss at 1.0858, eight pips above his entry — tight, because this is scalping. For most of the evening the spread is around one pip, so the bid and the ask are practically glued together. At 11pm the rollover begins. Within a dozen seconds the spread widens from one pip to twelve. The bid climbs to 1.0853 and stalls there — the chart shows that price never reached 1.0858. But with a twelve-pip spread the ask at that moment is 1.0865, clearly above the stop level. The buy order that closes the short position fills, and Marek exits at a loss even though his read on direction was correct.

Had Marek given the stop a margin of fifteen or twenty pips above the typical overnight gap, the same candle would not have touched his position, and in the morning he would have seen the rate moving his way. The entire difference between a loss and an open profit was a few pips of buffer, missing in exactly the place where the market is least predictable.

Then there is swap, and on Wednesday a triple one

The rollover window is not only about the spread. It is also the moment when the broker charges the swap point for holding a position overnight. If the interest rate differential works against your position, the swap is negative and you simply pay a financing cost. The MetaTrader 5 documentation describes this directly: at the point a position is moved to the next trading day, the system books the swap. The mechanics of that charge are laid out in more detail in the article on the forex swap.

There is also a calendar catch. The currency market settles trades on a two-business-day basis, so a position held through the Wednesday rollover carries the coming weekend on its value date. For that reason the swap is charged triple on Wednesday evening. For a position with a negative swap point that is a tripled cost in a single night — one more reason to treat the Wednesday window with more care than the rest.

It is natural mechanics, not a broker conspiracy

This is worth saying plainly, because the emotion after a stopped-out trade can bury the facts. The spread widening at midnight is, in the overwhelming majority of cases, not manipulation. It is the natural effect of vanishing liquidity multiplied by the mechanics of two prices. An agency broker receives a wider spread from its providers and passes it on, because that is what it is being quoted. The scale and the exact time differ between brokers, between instruments and between seasons.

That said, not every broker behaves the same way. Some on a market-maker model can widen the quotes in this window more than the market would justify — and that can be an abuse. So a measure of scepticism is healthy, but aim it at the right question: not "did they rob me", but "is the gap at my broker typical of the market, or suspiciously wide". The cleanest way to check is to compare how the spread behaves at two different brokers at the very same minute. The mechanics of deliberately running stops I cover separately in the piece on stop hunting.

What to do tomorrow

1. Measure the real spike time at your own broker. Over the next three or four evenings, watch the EUR/USD spread between 10:30pm and 12:30am and note in a simple sheet which minute it widens at and by how much. After a week you will have your own concrete number instead of a general theory — and that number, not a forum legend, should drive your decisions.
2. Give your stops a margin above the typical overnight spread. If your measurement shows the gap reaching twelve pips, set the stop loss with at least a dozen extra pips of buffer above that level for positions held through the window. A tight scalping stop makes sense in the middle of a liquid session, not at the turn of the day when the market breathes at its shallowest.
3. Do not open new short-term positions inside the rollover window. Entering just before 11pm means a worse execution price and a higher risk of slippage, because you are paying the widest spread of the whole day. If a signal appears at that hour, it is better to skip it and wait for the Asian open than to enter into the worst possible liquidity.
4. Consider closing scalping positions before the window, especially on Wednesday. On a strategy measured in minutes, holding through the rollover rarely pays — you risk a stop-out and you pay swap, triple on Wednesday. Decide consciously before 11pm whether a given trade truly needs to be held overnight, or whether you can close it cleanly and return in the morning.