Gold does not react to the number, it reacts to the rate

A hot CPI print does not move gold because inflation is higher. It moves gold because the market reprices the policy path, which changes real yields, which changes the opportunity cost of holding an asset that pays nothing. Everything in this lesson follows from that chain.

So the question before a data release is not "will CPI beat?" It is "how far does the market repricing push real yields, and how much gold does that imply?"

Beta, not guesswork

Gold's sensitivity to real yields is not constant, but it is measurable. Estimate it by regressing historical gold percentage returns on changes in the relevant real yield. That gives you a beta: roughly, how many percent gold moves per percentage point change in real yields.

Once you have that beta — and the market's implied move in real yields ahead of the event — you can compute an expected gold move. That number is the single most useful input into deciding whether to hold through a release.

Three decisions, in order

Expected move versus your stop. If the expected move is larger than your stop distance, you will probably be stopped out even if you are right about direction. That is not bad luck, it is bad sizing.
Size down, or stand aside. Halving size before a release is the standard professional response. It keeps you in the trade and halves the damage from an adverse gap.
Spreads widen. Your stop will be filled worse than you modelled. Add a slippage allowance to the expected move before you compare it to your stop.

What "holding through" actually means

Holding through a release is not a conviction decision, it is an arithmetic one. If your stop is inside the expected move plus slippage, holding through means accepting a high probability of being stopped at the worst price. If your stop is well outside it, holding through is reasonable and you get paid for the repricing.

How to check this yourself

Find the next scheduled inflation print or central-bank decision on an economic calendar. Before it, write down your expected real-yield change, then convert it — using the sensitivity you have observed in past episodes — into an expected gold move in dollars. Now check whether your stop survives that move. If it does not, the trade was never sized for the event, however right your view turns out to be.