Home › Gold forecasts › Saturday 3 October 2026
Gold closed the week at $4,140, down $145 (3.4%), a second weekly decline in a row. Two verdicts this week: Saturday’s call was wrong by ten dollars, and Tuesday’s reading was right in its first scenario. More important than either: we found a gap in our own record that had flattered us, and we brought the accuracy figure down from 96% to 90% ourselves.
Friday’s close $4,140 · Previous Friday $4,285 · 21K in Suez EGP 6,130 ·
Saturday 26 September: 45% consolidation ($4,230–$4,340), 30% a neckline break ($4,150–$4,230), 25% back above the floor ($4,340–$4,420). The close was $4,140.19, ten dollars below the lowest number. Wrong. We named the reason on Monday: we tied the week to US data and left out the oil channel, and the fall came through Hormuz before any data. The whole range was $270 wide in a market that moved $170 in a single day.
Tuesday 29 September: 40% a range below the broken floor ($4,050–$4,180), 30% further decline, 30% a bounce. $4,140 is inside the first. Right, and ranked first. As we wrote then: we do not edit an old forecast to make it right; we publish a new one beside it and grade both. The first was wrong and the second right, exactly as Friday morning’s “map for tonight” said for the $4,050–$4,150 zone.
Friday’s daily call: the close landed in the “strong payrolls” range ($4,090–$4,150), but payrolls were very weak (29,000 against 90,000). The confirming number (2-year above 4.85%) was not met; it closed at 4.84%. So, as promised on Thursday, this is correct by chance: it counts as correct under the rule, and we write “by chance” beside it. Gold rose to $4,227 on the print, then yields turned back up and oil rose on news of a third carrier group, and the market sold.
Correcting the record: in today’s review we found that the bucket miss of the 19–25 September week (close $4,285, below our $4,300–$4,420 range) was described at the time as “counting against us” but never entered the tally as a miss. That was lenient, and we fix it: both weekly calls now count as wrong. The true record: 31 published · 26 correct · 90% (26 of 29 graded; three misses: 10 September, the week of 25 September, the week of 2 October).
Saturday’s call → · Tuesday’s reading → · How grading works →
| Now | Reference | |
|---|---|---|
| Friday’s close | $4,140 | −$145 on the week |
| 24K, per gram, we sell | EGP 7,005 | down from 7,040 |
| 22K, per gram, we sell | EGP 6,421 | down from 6,453 |
| 21K, per gram, we sell | EGP 6,130 | down EGP 30 |
| 18K, per gram, we sell | EGP 5,254 | down from 5,280 |
📉 21K at EGP 6,130 (buy 6,070) from our counter reset at 10:30 AM today, down EGP 30 on Friday. World markets are closed until Monday. On the week the ounce fell 3.4% and the gram only 1.9% (6,250 to 6,130), because the dollar rate implied in our counter price rose from 51.85 last Friday to 52.63 today. Live prices are always at today’s prices.
Two weekly calls in a row failed, both with the same two faults. We will not settle for an apology. These are the rules we hold ourselves to from today, written down so you can hold us to them:
Kitco’s survey published on Friday 25 September read 5 up, 4 down, 5 sideways out of 14. Gold fell $145, so 4 of 14 had the direction. And us? Saturday’s call said “consolidation” (wrong on direction and bucket); Tuesday’s reading said “a range below the broken floor” (right). The plain result: we were half right, Wall Street a third right. Nothing to celebrate. The one difference: we grade ourselves by the number, and most names in that survey do not publish their result.
The new survey (Friday 2 October, 13 analysts): 6 bearish (46%), 3 bullish (23%), 4 neutral (31%). Main Street (182 votes): 47% bullish, 33% bearish, 20% neutral. Wall Street is close to a bearish majority for the first time in months.
Covering the Friday 9 October 2026 close · written under the new rules · total range $400
🇪🇬 21K in Suez at the end of next week: EGP 5,774 – 6,366, most likely EGP 5,966 – 6,144, assuming the implied dollar rate holds at 52.63.
✏️ 70% of our weight is below $4,150. This is a bearish call, said plainly, and it stands with the 6 of 13 on Wall Street. If it is wrong we will write that, as we did today.
The big lesson: a very weak jobs print, October odds at 20% and a soft PCE, and none of it lifted gold. What runs the market now is long-dated yields (debt worries and energy-driven inflation) and oil. Not the Fed.
| If the ounce is | 21K per gram works out near |
|---|---|
| $3,900 | EGP 5,774 |
| $3,960 (July low) | EGP 5,863 |
| $4,000 | EGP 5,922 |
| $4,050 | EGP 5,996 |
| $4,100 | EGP 6,070 |
| $4,140 (Friday’s close) | EGP 6,130 |
| $4,200 | EGP 6,218 |
| $4,279 (100-day average) | EGP 6,335 |
| $4,300 | EGP 6,366 |
📍 Raw metal at today’s local conversion factor (1.4806), before making charge, hallmarking and tax.
| $4,315 | The last bounce high |
| $4,280–$4,330 | Moving-average cluster (100, 50 and 20-day) |
| $4,200 | Sellers’ ceiling |
| $4,150 | The line that judges our call |
| $4,100 | First support |
| $3,996 | 29 July low |
| $3,960 | 17 July low |
The number that judges next week is $4,150: below it, our main call is right. A daily close above $4,200 makes the bounce serious. Below $3,996 the market is testing the summer low, somewhere it has not been since July.
Monday 5 October — 5:00 PM: ISM services (expected 55.7).
Tuesday 6 October — trade balance; 3-year note auction.
Wednesday 7 October — 9:00 PM: the Fed minutes (of the September
hike meeting); 10-year note auction, which matters for yields.
Thursday 8 October — 3:30 PM: jobless claims (expected 195,000); 30-year bond auction.
Friday 9 October — 5:00 PM: Michigan consumer sentiment; midnight: the weekly
close our call is graded on.
All week: Hormuz (a third carrier group; talks through Qatar) and Fed speakers.
Daily reports resume Monday morning, same rule: the number before it happens, the grading after.
Session closes, payrolls, unemployment, wages, Treasury yields, oil, hike odds and the Jefferson remark from Trading Economics; Kitco’s weekly surveys of 25 September and 2 October 2026; technical levels (July lows, the moving-average cluster, $4,315) from FXStreet; the carrier-group and Marines reports from Trading Economics; next week’s calendar from Trading Economics; local prices from our own published price list, reset at 10:30 AM on 3 October.
This report is the English edition. The Arabic edition is primary and was published first, at the same date’s Arabic page. Every figure here appears there.