London fix vs New York close: what each one does to your chart

They sit six hours apart and get mixed up all the time, but they are different kinds of thing. The London fix is a price that gets calculated: a benchmark rate set at 16:00 London time that funds and companies trade against. The New York close is a line on the calendar: at 17:00 New York time one trading day ends and the next begins.

Below is what each one is, when it happens on your clock, and what nine months of our own one-minute data show each one doing to the chart.

Each one in one sentence

What the London fix is

The fix is a reference rate. Funds and companies ask their banks to buy or sell currency at whatever the fix turns out to be, so a large block of orders lands on the same few minutes every day. We touched on it in our guide to the New York session; this post goes further.

The calculation window used to be one minute, centred on 16:00. After the scandal in which numerous banks were fined for their trading around the fix, a Financial Stability Board group recommended widening it. Guy Debelle of the Reserve Bank of Australia described the change at the time: WM moved to a five-minute window from 15 February 2015, the idea being that a wider window is harder to push around.

Month-end is when the fix matters most. Melvin and Prins used the fix to pin down when international equity funds trade their currency hedges. They found that when a country's stocks rise over the month, its currency tends to fall before the month-end fix, which they read as evidence that the hedging moves exchange rates.

And the "New York fix"?

There is no New York twin of the London fix playing the same role. When traders say "New York fix", they usually mean one of two things:

• The New York close, 17:00 New York. This is where the trading day ends. FXCM's help page on rollover puts it plainly: positions open at that moment are held overnight, and Wednesday's rollover books three days of swap because of the weekend.
• The New York option cut, 10:00 New York. It is a standard expiry time for currency options, and market news services publish each day's large expiries for it, as in this investingLive note for 21 September 2026. When a big expiry sits close to the price, it can keep spot contained around that level until the cut.

Neither is a calculated benchmark like the London fix. This post compares the fix with the close, because those are the two that mark every trading day.

The times on your clock

The UK and the US change their clocks on different dates, so the gap between the two moments is not always six hours: for a few weeks a year it is five. Many MetaTrader 5 (MT5) brokers, including the one that feeds our data, run their servers on Eastern European Time (EET, UTC+2) in winter and Eastern European Summer Time (EEST, UTC+3) in summer, which puts their midnight on the New York close.

In the second and last rows, the New York close comes an hour earlier on a broker clock.

What our data shows

We used one-minute bars for ten pairs (EUR/USD, GBP/USD, USD/JPY, USD/CHF, USD/CAD, AUD/USD, NZD/USD, EUR/GBP, EUR/JPY and EUR/CHF) from 12 January to 7 October 2026: 193 trading days, nine of them month-ends. The bars come from the broker that feeds our MT5 bridge and show the bid price.

For each pair and day, we took the six one-minute bars around a round hour (from :57 to :02) and compared them with the same window at other hours of that day. Two numbers came out: how much price moved, and how many ticks (price updates) there were. A value of 1.00 means an ordinary hour.

The 15:00 London window is left out of the comparison: for most of the year it is also 10:00 in New York, when many US releases and the option cut land. In the three weeks of March when the clocks were out of step, it was 14:00 London that fell on 10:00 New York, which may lift the 14:00 figure a little.

Medians across 1,866 pair-days, month-ends included (90 of them). Each value is the ratio to the other hours of the same day and pair.

On an ordinary day, the fix stands out more for activity than for movement. Leaving month-ends out, it has 60% more ticks than a normal hour, more than any other hour, and price moves 29% more, against 18% at 14:00 London.

With a narrower window of five bars (from :58 to :02), 14:00 even comes out ahead: 1.21, against 1.17 for the fix.

All ten pairs show the same pattern. Across all days, price movement in the fix window ranges from 1.14 to 1.50 times normal, and ticks from 1.48 to 1.79.

At month-end it is a different event. Price moves about twice as much as at other hours (2.06, interval 1.68 to 2.85) and ticks are 2.5 times normal. With only nine month-ends, read the size loosely; even the low end of the interval, 1.68, sits well above the 1.29 of the other days.

Minute by minute

The jumps come at 15:55 and in the 15:57 bar, as the window opens at 15:57:30, and again in the 16:02 and 16:03 bars, as it closes at 16:02:30 and just after. In between, from 15:58 to 16:01, there are more ticks than in the two hours before, but price moves about as much as it did then. Other hours show a smaller bump just before the hour too, so not all of it is the fix.

That quiet middle is consistent with what Ito and Yamada found after the 2015 reform: banks changed how they execute fix orders, and the volume spike inside the window disappeared, although the price anomalies around it did not. Ticks are not volume, though, and ours are still above normal inside the window.

A month-end fix: EUR/JPY on 30 September 2026

The last day of September was also a quarter-end. EUR/JPY was at 178.093 at 15:53 London, and at 178.474 by 16:00, 38 pips higher in seven minutes.

USD/JPY rose 20 pips over the same minutes, so it was mostly the yen being sold. By 16:10 EUR/JPY had given back six and a half pips, to 178.409.

Does the move into the fix reverse?

Martin Evans, looking at 21 currencies over a decade, found extraordinary volatility and negative serial correlation around the fix: what went up into 16:00 tended to come down after. We checked the same thing: the move from 15:30 to 16:00 against the move from 16:00 to 16:30.

Across all days the correlation is −0.12 (interval −0.23 to −0.01), within the −0.18 to +0.08 of the other hours. Leaving month-ends out it is −0.09 (−0.21 to +0.01), again inside the other hours' range (−0.12 to +0.09). At month-end it is −0.48, with an interval from −0.77 to +0.22 that crosses zero.

Nine months and nine month-ends cannot confirm or refute a decade-long study. What they do rule out, on ordinary days, is a strong reversal: the correlation is no stronger than −0.21.

The New York close: the bid falls, then comes back

At 17:00 New York, the picture flips. Ticks in the 17:00 bar drop to less than a fifth of that afternoon's normal, while the bar's move is about 17 times the afternoon's average minute.

That is not a burst of trading: there is almost none. It is the spread opening up. Our bars show the bid, and as liquidity thins at the turn of the day, the bid drops away from the ask.

Measured from the price at 16:59 to the low of the 17:00 bar, the median bid drop is 7.6 pips. At 16:00 and 18:00 New York the same measure gives 0.7. By pair:

Monday to Thursday only: a Friday 17:00 is the weekend close, which is the weekend gap's story.

Most of it comes back fast: by about 17:15 the bid is a median 5.8 pips above that low. An hour later it is still a median 1.7 pips below where it was at 16:59, which may mean the spread stays wider than in the day; we cannot see the ask, so we cannot measure the spread itself.

On about one in ten Monday-to-Thursday pair-days there is no 17:00 bar at all: no price update was recorded in that minute.

On 6 October 2026, for example, EUR/JPY was at 178.014 at 16:59 New York. The 17:00 bar traded down to 177.855, about 16 pips lower, on 23 ticks, and by 17:10 it was back at 178.023.

That was not a wave of selling: for a minute, there were simply few quotes nearby. This is the same mechanism as the overnight spread, compressed into a few minutes.

What should you do with this?

• Don't read the fix as a signal. On an ordinary day, 15:55 to 16:05 London is busy, not directional. At month-end and quarter-end, expect bigger moves in both directions and keep tight stops out of the way.
• Mind your stops across 17:00 New York. A long position's stop-loss is triggered by the bid, and the bid is exactly what drops. A stop a few pips under price on EUR/JPY or EUR/CHF can be hit by a spread, not by the market. The ask, which our data does not show, should widen the other way, so a short position's stop is exposed too. Our guide to stop placement has more on giving a stop room.
• Ignore the 17:00 wick. On one-hour (H1) charts and lower, the long lower wick at the New York close is often the spread, not a price anyone traded at in size. Don't mark it as a level.
• Check your own broker. These are one broker's numbers. Open a one-minute chart at 17:00 New York tomorrow and look at your own spread.

What this does not tell you

It does not tell you where the fix rate itself was set: we have our broker's prices, not WM/Reuters'. It does not measure the ask, so the spread at the close is inferred from the bid, not measured.

It is nine months of one broker's feed, with nine month-ends, which is why the month-end figures have wide intervals. And ticks are a broker's price updates, not traded volume, as our comparison of volatility, liquidity and volume explains.