The theme. Tech under pressure again over doubts about the costs of artificial intelligence; oil climbing back with the attacks in Saudi Arabia; gold below 4,000.
The method. This week no orders. Gold and dollar/yen already have their move in play, today both against; the Nasdaq, the only index measured, is rejected by the numbers; oil, the briefing writes, has no solid level either below or above the price.
To take home. Admitted does not mean order. And an explanation found after the move is not a measurement.
The past week (19-23 Oct)
Tech. Thursday 22 Oct Nasdaq 100 −2.9%, semiconductors (SOX) −6.7%, S&P −1.2%: the worst day since July. Meta −7% after its results, with 2026 investment raised to 135-145 billion.
- The why: whoever raises spending promises less certain profits. If the market doubts that this spending, partly made with debt, will pay for itself, it lowers the price today: Reuters calls it the fear over the costs and the debt of artificial intelligence.
- The example: the broker's Nasdaq 100 from the high of 6 Oct, 31,365.74, to Friday's close, 28,993.39: −7.6%. The Dow, where tech weighs less, closed the week at +0.4%.
Rates. US ten-year from 5.09% to 5.00% (−9 bp, basis points: hundredths of a point); on 22 Oct alone −17 bp. The why: when the stock market falls hard, money looks for government bonds, their price rises and the yield, which moves the opposite way, falls.
Oil. Down early in the week on the diplomacy over Hormuz (the CFD from 99.24 on Friday 16 to 97.53 on Wednesday 21), then up with the Houthi attacks in Saudi Arabia: on Thursday Aramco evacuated its Dhahran headquarters. Friday Brent at 116.72 (+3.1%); December WTI, the one the CFD follows, +3.8% on the week, at 103.
- The why: the price holds a premium for the risk of losing supply. Every attack near the wells raises it, every diplomatic opening lowers it. And little gets through Hormuz: 4 transits on 18 Oct, against 85 a day before the crisis.
- Expensive oil means more inflation, and a Fed less free not to raise rates.
A stronger yen. Dollar/yen −0.7% at 156.8. BoJ governor Ueda early in the week and deputy governor Himino on 21 Oct («do not postpone the adjustments») spoke like hawks, that is like those who want to raise rates. If Japan raises, the gap with US rates narrows, holding dollars against yen pays less and the yen rises. September core inflation came out at 2.1%, the expected number written here last Sunday: according to the sources the words moved the market, not the data.
Gold −4.0%, below 4,000 (3,995, low 3,975 on Friday); silver −5.3%.
Question. A frightened stock market, rates down, and the December Fed hike priced in falling during the week from about 89% to 75-80%: usually all this helps gold, which pays no interest. So why −4.0%?
The sources' explanation is expensive oil, which keeps rate expectations high: it holds only in part, because those expectations stay high but have come down. The «forced liquidation» (those losing elsewhere unload what they have to cover margins) is an opinion of Checkan (Asset Strategies), not a fact. The fact is that on Thursday gold, Bitcoin and the Nasdaq fell together. The lesson: an explanation found afterwards is a story, not a measurement. That is why the method does not ask «why»: it writes the stop and the invalidation first.
Europe and US data. On 21 Oct the French ten-year above 5%, the first time since 2002; eurozone flash PMI at a 4-month low. On Friday evening Moody's left France untouched. In the US, manufacturing PMI 56.0 against 55.5 expected, a 53-month high; prices charged rising at the fastest pace since August 2022; Michigan inflation expectations at 4.7% at 1 year and at 3.5% over the far horizon. Prices and expectations are exactly what the Fed cares about: whoever expects more inflation asks for higher wages, and inflation feeds itself.
The weekend. On Saturday a Houthi missile headed for Riyadh was intercepted, and a drone shot down in the Eastern Province. According to the Wall Street Journal the aircraft carrier George Washington is being rearmed with bunker-buster bombs; US envoy Tom Barrack said a strike «this weekend» is possible, while on 9 Oct Trump had written that he will not attack Iran before the 3 Nov election. In the news found by the briefing, no attack. The Iranian ambassador to Pakistan claims to have offered to reopen Hormuz, the Mehr agency denies it: until one version is confirmed, neither is a fact.
- For Monday: with the market closed the news piles up and the price discounts it all at once at the open, starting with oil. The first price can be far from Friday's (a gap), and a stop in between is closed at the first available price, not at its own.
The week ahead (26-30 Oct). Shifted times: Italy has gone back to winter time, the US goes back on 1 Nov. US data come out at 13:30 Italian time, Wall Street opens at 14:30 and closes at 21:00.
- Monday 26 Oct at 10:00 the German Ifo (expected 89.8); Tuesday at 13:30 US durable goods (+0.6%).
- Wednesday 28 Oct at 19:00 the Fed (press conference at 19:30), expected on hold at 3.75-4.00%; after the close Microsoft and Alphabet.
- Thursday 29 Oct at 13:30 US GDP (+3.2%) and September PCE, the inflation the Fed watches most (core +0.3% on the month, +3.1% on the year); at 14:15 the ECB, expected on hold at 2.50%; in the evening Amazon and Intel.
- Friday 30 Oct around 04:00 the BoJ, expected on hold at 1.25%. Sunday 1 Nov OPEC+ on December output.
What is already priced in. Fed on hold at about 90% (hike at 10% on FedWatch); in December a hike at 75-80%. ECB on hold at 98%. BoJ: hike below 5%; for December, 80% on swaps but 52% on futures, the sources do not agree. An event priced at 90% moves little if it happens: what moves is the surprise, and what the central bank lets on about the next meeting.
If… then… (on the weekly moves in play)
- Gold (stop 3,940, Friday 3,995.43): if on Wednesday the Fed hiked (priced at 10%), usually rates and the dollar rise and gold falls, towards the stop: it is among the risks in the record. If it stays on hold and takes the urgency off December, usually the opposite.
- Dollar/yen (stop 155.550, Friday 156.827): a BoJ hike would be a full surprise, yen up and price down towards the invalidation at 155.60. A surprise Fed hike would do the opposite.
Why the weekly exists. The average move grows with time, the noise with the square root of time: the first is worth 2% of the second at 3 hours, 15% at 20 sessions, 26% at 60. With the stop at 2 ATR (two average days) the long side beats the cost on 8 instruments out of 14, the short side always loses: that is why the weekly is long only. It is the risk premium, plus the carry where the broker pays the swap. It is not a forecast: it holds while markets rise, and in a three-year decline it loses.
- The example: the FTSE 100 over 20 sessions rises on average by +0.188 ATR, but spread and swap cost 0.439. Edge -0.252: out. Rising is not enough, it has to rise more than it costs to hold.
Admitted does not mean order: the Nasdaq. It has the highest edge in the field, +0.579 ATR, and it has fallen 7.6% from the high: it looks like the right moment. The method measures what happened after a drop like this: after more than 5% in 20 sessions, over the next 20 it returned +0.129 R, against +0.235 usually. Minus the cost, 13.2% of the stop, what is left is −0.003: zero. And it is UNSTABLE: over the last 3 years the same as usual (+0.013 over 47 cases), and a third of the cases come from 2022. It does not mean it will fall: it means the idea does not pay its costs.
- A 5-session variant would make +0.251 net, but it is a rebound, a different trade: changing the question until you like the answer is the simplest way to fool yourself.
- Last week's limit was the same, 28,800: on Friday the price fell to 28,806.16 and came back up, expired by 6 points. Moving it higher would change which trades are taken: it has to be measured first on all the cases, not decided afterwards on one.
The field and the swap (reading of 23 Oct 2026 at 22:59, compared with 16 Oct). The same 8 out of 14 admitted: S&P 500, Nasdaq, Dow Jones, Nikkei, gold, silver, WTI oil, dollar/yen. On the long side the swap is earned only on WTI oil (+0.0129 ATR a night) and dollar/yen (+0.0083). Only oil's has changed, by 101%: today it is 66% of its edge, and it can turn. Dollar/yen over the last 3 years has moved by +0.027: about half of its edge is the carry.
Already in play (MetaTrader daily bars)
- Gold, 2 Oct: limit 4,105 filled on 6 Oct; low 3,975.61 on 23 Oct, Friday 3,995.43: $110 against, with the stop at $165. Confidence LOW, 0 out of 6. Time exit on 2 Nov 2026.
- Dollar/yen, 16 Oct: limit 157.680 filled on 22 Oct; low 156.068 on 23 Oct, Friday 156.827: 85 pips against, with the stop at 213 pips. Time exit in the record: 18 Nov 2026.
A long limit fills when the price falls, and it can keep falling afterwards: that is why stop and invalidation are written first. The daily has 8 moves in play: the teacher covers them from Monday.
Brake and track record. 5 weekly proposals: 3 not filled, 2 in progress. Sum of net R +0.00; the brake triggers at -12 R and has not triggered. On top of that, 1 rejected, which the record follows all the same: the «no» is judged afterwards too.
The orders of the week. These are the method's orders, registered before and judged after in the public record: they are not financial advice.
This week none. A single line, with the server date: Friday 23 Oct 2026 at 22:59.
- Nasdaq · long · REJECTED 2026-10-23-01: measured thesis, not an order. After a drop like this it returned less than usual, and holding it 20 sessions does not pay the costs (net R −0.003, confidence NONE, 0 out of 6).
Summary: no new orders; in play the gold of 2 Oct and the dollar/yen of 16 Oct.