Gold at 4,045.80: mid-range, undecided

Both ends of the range are still unresolved, so the next session inherits an open argument. Against the 10-bar mean of 4,072.89 price is -0.67% away; average daily range over the same window is 87.57, against today's 26.02. Wide deviation with a contained range reads as an orderly trend rather than a squeeze, and the two need opposite handling: one you sit through, the other you stand aside from.

Asia 4,065.44 to 4,045.80 (-19.64, 147 prints); Europe 4,105.98 to 4,094.41 (-11.57, 18 prints); US 4,123.73 to 4,033.19 (-90.54, 136 prints). Net move for the session is 12.50, so the largest window carries 724% of it.

Read the share, not the sign. US carried 90.54 while the other windows mostly held the line - buying was a decision taken inside one window, not demand accumulating through the day. Real accumulation stairs the closes higher; this was one leg and then a lot of sideways.

The last print sits 6.77 under the high instead of at it, and that gap is the only bearish line in this section. Same range, a close on the low, and the read flips: the buying would have been absorbed. It was not, so the burden of proof moves to sellers at the next open.

Second thread: economic data. 2 wires in the window sit under this heading, the strongest filed at 22:30 UTC. The feed is Chinese-language, so what carries here is the category and its timing rather than a translated quote; the links sit in the sources section for anyone who reads the original. It supports the same conclusion without carrying it alone.

Also in the window: flows and positioning. 2 wires in the window sit under this heading, the strongest filed at 22:30 UTC. The feed is Chinese-language, so what carries here is the category and its timing rather than a translated quote; the links sit in the sources section for anyone who reads the original. It supports the same conclusion without carrying it alone.

Cross-check through the dollar: EUR/USD +0.03%, GBP/USD +0.04%, USD/JPY +0.05%. Two of three line up, so the dollar move is partial and gold is partly trading its own book. Gold rising into that is the result, not the cause.

The falsification test is what keeps this section honest. A print that should have pressured gold and did not is not the driver - the tape has already voted on it. Calendar and wires above were checked against the price path, and where they disagree the path wins, because it is the only witness that cannot edit what it did.

Extension. The same range projects 4,222.41 at 1.272 and 4,317.66 at 1.618. Price at 4,045.80 is 176.61 short of the first extension, so that level is overhead supply to reclaim, not a floor underneath.

Stretch. Price is -0.67% from the 10-bar mean of 4,072.89, today's range 26.02 against an average 87.57. Deviation this wide with a range this contained means the trend is paid for in instalments rather than one repricing: the kind that continues but does not pay you for chasing it.

The longer frame. Against the 20-day at 4,124.82 price is -1.92% away; against the 50-day at 4,268.38 it is -5.21%. Below both means, so rallies are still being sold into.

Momentum. RSI(14) reads 41.4, neither stretched nor washed out. MACD sits at -47.04 against a signal of -56.26, histogram +18.44 - a cross above the signal line but still under zero, which historically marks a bounce rather than a new leg. Twenty sessions have moved -2.98% in total, average range 81.85.

How to use the list: the number matters less than the reaction. A strong print that gold shrugs off says more about positioning than the print did, and that reaction - not the headline - is what carries into tomorrow.

Path two, 38%: Digest between 4,222.41 and 4,009.88. Trigger: three sessions closing inside that band with the range contracting. Digestion is the healthiest outcome here - it repairs the stretch without breaking structure.

Path three, 26%: Lose 4,009.88 and the trend breaks. Trigger: a daily close below 4,009.88. This path needs no story to happen, which is exactly why it gets a weight.

These weights are the desk's read of the structure above - not a forecast, not a promise. They are written down for one purpose: so that tomorrow you can check whether the market did what the structure implied.

Every derived figure - range position, session shares, Fibonacci levels, the 10-bar mean, the extension overshoot - is computed from the series inside this file, not typed by hand. Recompute it and you should land on the same numbers; if you do not, the file is wrong and this section is the audit trail.

Known limits, stated rather than hidden: five-minute k-lines are not retrievable from the vendor for any period, and daily history is still accumulating, so no MA20/50 or momentum work appears. The intraday sequence depends on the poller's rolling window, which is archived to disk every ten minutes so that a restart cannot erase the day.

This is market commentary for traders, not investment advice: no position sizing, no entry instruction, no promise of outcome. If a number here disagrees with your own chart, trust your chart and tell us - a correction is worth more than agreement.