Part 2: The sessions · Chapter 4 of 14 · 9 min

London and the LBMA fixings

London is the historic centre of gold, but in our data its session does not set the extremes more than chance (54% of days against 51%), the 08:00 opening is not a peak (1.2× the median) and the 10:30 and 15:00 fixings leave no visible trace at five minutes. What sets the extremes of the day is New York: 82% of days, against 55% by chance.

Why London matters (and what the fixing is)

London is the centre of the physical gold market: the big banks keep their vaults here, bar contracts are settled here. When people say "the gold price" in the official sense they often mean the LBMA Gold Price, set twice a day (at 10:30 and 15:00 London time) by an electronic auction: participants enter buy and sell orders until they meet at a single price, which becomes the reference for the day.

Banks, mines, refiners and jewellers use that price to settle physical gold contracts. From this comes the belief that the fixing is a moment of tension for the market. It is reasonable to ask whether price moves more at 10:30 and 15:00. An auction, however, looks for the price the market is already making: it does not decide where it goes next. Let us check what the data show.

The 10:30 fixing: nothing

Take the five-minute candles around 10:30 London time (average range divided by the overall median):

London time10:2010:2510:3010:3510:40
Range relative to the median0.86×0.92×1.05×0.95×0.88×

The 10:30 fixing is worth 1.05× the median: a normal value, barely above its neighbours. The median move in the 30 minutes after 10:00 is 4.2 $, in the 30 minutes after 10:30 it is 3.8 $, against 5.0 $ after 09:00 and 4.8 $ after 12:00. Price does nothing visible around 10:30.

The 15:00 fixing: inside the American data

The 15:00 London candle is worth 2.21× the median. It looks like a peak, but 15:00 in London is 10:00 in New York, the time of the American 10:00 data. The two events almost always coincide. A natural experiment: in the three weeks of misalignment between America and Europe (15 days) 15:00 in London is 11:00 in New York. If the fixing made the difference, that candle should stay high while the data one falls. The opposite happens: London 14:00 (= New York 10:00) is worth 2.75×, London 15:00 (= New York 11:00) 2.40×, and the 09:30 New York candle 2.76×. The whole American morning is high on those days. There is no separate effect of the fixing: with 15 days the sample is small, but no signal tells it apart from the rest of the American session.

The London opening: a peak that is not there

The 08:00 London time is worth 1.2× the five-minute median, against 2.5× for Shanghai and 2.3× for the American data (chapter 2). We also check the most widespread idea, that London "reverses" or "continues" the Asian move. The first two hours of London (08:00-10:00) go in the same direction as the Asian session (00:00-07:00 UTC) on 46% of days (±7 points): a coin toss. The median move of those two hours is 9.1 $.

Who sets the high and the low of the day

The useful question for someone trading is another: in which session are the highs and lows of the day born? Gold's "trading day" runs from 18:00 New York time of the previous day to 17:00. Over 211 days we measure, for each session, on how many days the high or the low falls inside it, and compare it with a window of the same length chosen at random in the day (400 draws).

Window (local time)HoursMedian share of the rangeOne of the two extremesRandom window of the same length
Tokyo 09:00-15:00652%50%42% (from 37% to 47%)
London 08:00-16:00865%54%51% (from 45% to 56%)
New York 08:00-17:00968%82%55% (from 50% to 60%)
London "killzone" 02:00-05:00 New York329%10%23% (from 18% to 28%)
New York "killzone" 07:00-10:00 New York345%25%23% (from 18% to 28%)

The conclusions are clear. New York sets the extreme of the day in 82% of cases, well above the 55% of chance. London sits in chance (54% against 51%). Tokyo is just above (50% against 42%). The London "killzone", the 02:00-05:00 New York window that some schools indicate as the moment when the high or the low forms, in our data contains the extreme of the day only in 10% of cases, under the 18% that is the low limit of chance. The New York one (07:00-10:00) sits in chance.

A note: the windows overlap, and the American session is the longest and arrives at the end of the day (17:00), so it is partly favoured by the definition. Even with this caution the distance from chance is the largest in the table.

Figure 1. 20 February 2026 (UTC hours): the low, 4,985.60, was born at 03:00 in the Asian night; the high, 5,110.98, at 21:00 in New York. The grey and blue bands are the London and New York sessions; the two fixings fall in the London session and, at 15:00, also in the American one.Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 19 Feb 2026 23:00 UTC to 20 Feb 2026 21:00 UTC.
Read the chart in words

H1 gold chart from 19 to 20 February 2026 with the London and New York sessions highlighted and the two fixings marked.

WhenWhatPrice
20 Feb 2026 08:00 UTC → 20 Feb 2026 15:00 UTCLondon 08:00-16:00
20 Feb 2026 13:00 UTC → 20 Feb 2026 21:00 UTCNew York 08:00-17:00
20 Feb 2026 10:00 UTC10:30 fixing5,039.83
20 Feb 2026 15:00 UTC15:00 fixing5,067.68

Times in UTC.

What to do with it

  1. Do not treat the fixing as an event. At 10:30 price does nothing different; at 15:00 what counts is that the American data arrive. A reference auction for physical gold is no reason to enter or to stay out.
  2. Do not wait for the London opening as an event. For gold it counts less than Shanghai and much less than the American data. If your plan depends on "what happens at 09:00 Rome time", test it.
  3. Look for the highs and lows of the day after 14:30. The extremes form in the American session on 82 days out of 100. For strategies that want to enter near the extreme, the useful window is that one, not London.
  4. Treat "killzones" as hypotheses to be tested. The London one (02:00-05:00 New York) contains an extreme of the day on 10% of days, less than chance. Ten months and a single instrument are not enough to say they do not work at all, but they are enough not to build a system on them without having tested them.
  5. Remember what we measure. Here we measure where the extremes are born, not whether it is possible to make money trading in the windows. We test that in chapter 11.

Procedure in four steps

  1. Stop looking for the signal in the fixing minutes: it is not there.
  2. To look for the extremes of the day, look at the American session (from 14:30 Rome time).
  3. Before using a "killzone", measure on how many days it really contains the extreme and compare with a random window.
  4. Always check which local time you use (London, New York) and whether your chart converts it with daylight saving.

Exercises

Chapter exercises

1. On the chart the grey and blue bands are the London and New York sessions of 20 February 2026 (UTC hours). Which letter is the candle that contains the 10:30 London fixing (the 10:00-11:00 candle)?

Chart exercise. Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 19 Feb 2026 23:00 UTC to 20 Feb 2026 21:00 UTC.
Read the chart in words
WhenWhatPrice
20 Feb 2026 08:00 UTCPoint A5,033.49
20 Feb 2026 10:00 UTCPoint B5,039.83
20 Feb 2026 15:00 UTCPoint C5,067.68
20 Feb 2026 14:00 UTCPoint D5,064.23

Times in UTC.

2. The five-minute candle at 10:30 London time is worth 1.05 times the median, its neighbours 0.86-0.95. What does it mean for someone trading?

3. On how many days out of 100 is the high or the low of the day born in the New York session (08:00-17:00), against how many by chance?

4. The London "killzone" (02:00-05:00 New York) contains an extreme of the day on 10% of days, against 23% for a random three-hour window. What do you conclude?

Have a question or want to share your exercise?

Post in the community, or join the free signals room on Telegram.

Educational content, not financial advice. Trading involves risk.