Gold at 4,000: The Iron Floor That Wasn’t

In March, someone said 4,000 was an iron floor. The data was laid out neatly, table after table: historical ranges, central bank gold buying, geopolitical risk, dollar credit. Layer upon layer, until it all came down to one sentence: buy the dip in gold at 4,000.

It sounded like a wall. The market never follows the wall’s blueprint.

This week, gold has been squeezed from both sides. The US-Iran situation has not fully eased. Shipping risks remain. By all logic, safe-haven demand should be present. But high oil prices push inflation up, and if inflation will not come down, the Fed will not easily loosen its grip. Gold wants a bite of the safe-haven meal. Oil prices have brought the rent collector, interest rates, to the table.

So gold does not rise just because fighting breaks out. The market is calculating several ledgers at once: safe-haven demand, inflation, rates, the dollar, and flows. Price cannot depend on only the one entry you like. Explaining everything with geopolitical risk is like looking at a three-dimensional chessboard with one eye.

The first ledger is the Fed. This time the policy rate was held at 3.75% to 4.00%, with ten votes in favor and two votes arguing it was not tight enough. Those two wanted another 25 basis points. The dissenting votes did not mean “cut now.” They meant “with inflation like this, it still needs to be restrained.” The US economy has not clearly lost momentum. Core PCE is still at 3.1% year over year. The basis for expecting the Fed to rush out and lift prices is not sufficient.

A pause only means not moving for now. It is not a guarantee that the next move will go the way you want. Whether there is a hike in December still depends on the data. Market probabilities change. You cannot take an 80% probability at one moment and write it into the meeting minutes in advance. Probabilities are alive, meetings are held, and data arrives later. Mixing the three together creates the illusion that the answer is already known.

The second ledger is US Treasuries. The 10-year nominal yield is still near 5%, and the real yield is around 2.6%. Gold pays no interest. That is the opportunity cost it faces. You hold gold waiting for appreciation. Someone else holds bonds and earns coupon income. Bonds carry price and exchange-rate risk. They are not a sure profit. Still, some capital asks a practical question: why buy a metal that generates no cash flow now, instead of holding an asset with income and waiting?

Gold needs safe-haven and allocation demand to answer that question. The answer cannot hold only in a long essay. It also has to hold on the screen.

The third ledger is the tape. In Thursday night’s session, both the dollar and Treasury yields pulled back, and gold still fell. Even when conditions improved slightly, it did not rise. Watch whether buying is strong. As for forced liquidations or big money fleeing, without position evidence, I will not tell those stories. The biggest danger in market analysis is reasoning backward from “it fell” to “someone ran,” and then from “someone ran” to “it will fall further.” The story is smooth. The evidence is thin.

Below 4,000, the first level to watch is still 4,000 itself. If it breaks below and can be reclaimed quickly, and later pullbacks still find buyers, that suggests this break may not have created sustained selling pressure. If it falls, rebounds, gets pressed back at 4,000, and keeps making new lows, then that threshold may turn from psychological support into short-term resistance. As for 3,900 and 3,800 below, for now they are round-number watch levels, not verified strong support. Round-number levels look like steps. Without volume and buying support to verify them, the steps may just be lines drawn on paper.

Gold has another way of leading people astray: using long-term reasons to backstop a short-term entry price.

Dollar credit, central bank reserve diversification, and geopolitical risk can be reasons to study gold. They explain why to allocate over the long term. They do not explain why this price must be bought today. Gold is not a dividend-paying company. You cannot take a relatively stable operating profit and calculate a clear dividend return for it. How much cheaper it has become must be judged against the environment, not just against your own cost basis. Cost is personal memory. Price is market consensus. The market will not come back and apologize because someone bought too high.

Being long-term bullish does not mean any short-term price is fine. Allocating to gold does not mean every drop deserves more allocation. These two sentences sound like common sense. In the market, they are the easiest to forget. Once people turn long-term logic into faith, they explain every drop as an opportunity, then explain every opportunity as a reason to add. Research becomes a way to justify your own position.

So I read this loss of 4,000 as follows: the earlier repair was not stable enough, and the market needs to keep looking for support. First, see whether it can reclaim 4,000. Then see whether real rates, the dollar, and shipping conditions are cooperating. If it cannot reclaim 4,000, do not rush to issue 3,900 a “golden pit” certificate. Whether a pit is a pit is known only after a foot steps in and someone pulls it up. Based only on a round number and a long-term narrative, the certificate is issued too early.

A more practical approach is to break the question apart. The variables are whether 4,000 is reclaimed, whether pullbacks find support after a reclaim, whether real rates have pulled back, whether the dollar is merely rebounding or strengthening again, and whether shipping risk is building or only living in headlines. If these variables do not cooperate, the two words “safe haven” cannot support a sustained move.

With cash in hand, you can wait. With a full position, you can only hope.

Waiting is active. Hoping is passive. Those who wait look at conditions. Those who hope look for miracles. Gold has a long-term story. What it lacks is evidence of short-term support. Treat 4,000 as a long-short ledger, not an iron floor: one entry for safe-haven demand, one for inflation, one for rates, one for flows. Before the ledger balances, do not turn any single entry into a conclusion.

We respond. We do not pray.