Monday's Asia session, spot gold traded down to 3,978.01 and turned. By the New York session it had taken back 4,000 and was last at 4,021.75, up 0.66% on the day. That is the whole move. It is not a very dramatic move. But the location matters more than the size.
The zone I flagged on Sunday
In Sunday's edition I wrote that 4,000 had turned into resistance overhead and that 3,970 was the one buy-side zone that had not been tested yet. Monday's low was 3,978. It got within eight dollars of that level and then bounced.
I want to be precise about what that means, because it is easy to turn this into a story it isn't. The market did not touch 3,970. It came close and left. That is not the same event as testing the level and holding it, and I am not going to retroactively claim the call was confirmed.
Why the near-miss is more useful than the clean hit
Here is the part I actually find interesting. A level that gets approached and abandoned tells you something a level that gets touched squarely does not: it tells you someone was willing to act before the obvious price. When you see a market turn eight dollars short of a widely watched area, the useful takeaway is not "buy there" — it is that the area is now on the record as defended, and that the next approach to it will be a different situation than the first one.
This is the kind of thing I write down and date, so that three weeks later I can check whether I was right about the process, not just the outcome. Most traders only remember the outcome.
What the data actually says going into Wednesday
Numbers, not impressions:
- CME FedWatch: probability of at least one hike by the December meeting at 76.2%, up from 74.2% on Sunday.
- COT: managed-money net long in gold at 104,206 contracts, down 2,883 on the week. Positioning is being trimmed into the event, not added.
- US 10-year: nominal yield 5.00%, real (TIPS) yield 2.66%. Gold is priced off the real number — that is the one to watch, and it barely moved.
- Germany IFO: business climate 90.2.
- Brent: December contract at 115.15, down 1.35%.
The Fed is in its blackout period, so there is no new guidance coming from officials. Wednesday brings the decision; Thursday brings GDP and core PCE. That is two separate event risks inside 24 hours, and it is why I do not treat this as one trade.
What I do in a week like this
Not a prediction — a checklist I run every time, so the decision is made before the number prints rather than during:
- No new positions in the 24 hours before a decision. Not because I know the direction. Because I don't, and because the spread and slippage in that window are a cost I can simply decline to pay.
- Write the level down before the event, with the date. 3,970 tested-to-8-dollars on 27 Oct. If I revisit this in November I want to know what I actually said, not what I remember saying.
- Separate nominal from real yields every single time. A headline that says "yields up, gold should fall" is usually reading the wrong number.
- Check whether positioning is being added or trimmed. Managed money down 2,883 into an event is a different setup than up 2,883, even with an identical price chart.
The honest version
I have been trading this system for ten years and something over 18,000 trades. The reason I keep a written process is not that it makes me right more often — it is that it makes my mistakes legible. A trade that loses for a reason I wrote down in advance is a data point. A trade that loses because I improvised during a press conference is just a loss.
Monday gave me a near-miss at a level I had marked, in a week with two event risks stacked. That is not a signal. It is a note to myself, and I am publishing it so you can hold me to it.
Levels and structure here are observations for study. The Fibonacci tools I use are analysis indicators, not signals — nothing in this note is a recommendation to buy or sell.