Gold Went From $5,600 to $4,100 This Year. Your Settings Did Not Notice

Gold reached an all time high around $5,602 on 29 January 2026, and traded near $4,140 in early October. Whatever you think of the path, the level changed by roughly a quarter inside nine months.

Now consider a typical automated setup on gold. A stop loss written as a dollar amount. A grid or scale in distance written as a dollar amount. An entry filter requiring price to be some dollar distance from a moving average. All three were chosen at some price level, and none of them changed when the level did.

That is the whole article. The settings are still doing exactly what you told them. What they mean in relative terms is no longer what you meant.

A stop of eleven dollars is a different instruction at different price levels, because what matters to a strategy is not the dollar amount but how it compares to how far the instrument typically travels.

Instruments tend to move in rough proportion to their price. At a higher level the same dollar distance is a smaller share of a typical session's range, so a stop that used to sit outside the noise now sits inside it, and gets hit by movement that is not a signal. At a lower level the reverse happens: the stop sits further out than intended and each loss costs more than the plan assumed.

Neither failure announces itself. The strategy does not error. It simply starts producing a different distribution of outcomes than the one it was tested on, and the trade log looks like a strategy that stopped working rather than a setting that went stale.

The same drift applies to grid spacing, to trailing distance, and to any entry filter expressed in price rather than in proportion.

On a personal account this is an annoyance. On an account with a drawdown limit it is a mismatch between two things measured in different units.

Your firm's limits are percentages: a daily loss of some percent, a total drawdown of some percent. Your stop is a dollar amount. When the price level moves, the percentage stays fixed and the dollar stop silently changes what fraction of the limit a single loss consumes.

Work it through with your own numbers rather than taking ours. Take the dollar value of one full stop at your current lot size, divide it by your account size, and compare that percentage to your daily limit. That tells you how many consecutive full stops your day survives. Then do it again with the price level from when the settings were written. If the two answers differ meaningfully, the settings drifted.

The arithmetic behind that is in position sizing for a prop firm challenge, and what a daily limit actually measures against is in why most traders fail prop firm challenges.

This distinction is worth holding onto, because the two look similar and one of them is a trap.

Optimisation searches settings for the combination that performed best over a past period. It finds the values that fit that history most closely, and some of that fit is coincidence. Run enough combinations and you will always find one that looks exceptional, including for a strategy with no edge. The trap and the defence are in EA backtest against live results.

Recalibration is different and much duller. It restates the same intent at the new level. If the stop was originally chosen to sit outside a typical session's range, you measure the typical range now and restate it. You are not searching for a better setting, you are preserving the one you had.

A useful habit: when you write a setting, write down the reason next to it. A stop of eleven dollars with no note is a number nobody can recalibrate later. A stop that reads as roughly one and a half times the average session range can be restated by anyone, at any price level.

One operational detail that causes more confusion than it should, specifically on gold.

Gold is commonly quoted to two decimal places, and the engine fields in most expert advisors are expressed in points rather than in dollars. On a two digit quote, one hundred points is one dollar. A setting of 1100 is eleven dollars, not eleven hundred.

Get that wrong in either direction and the strategy is unrecognisable: either it never triggers, or every stop is a hundred times wider than intended. We carry a note about this in our own files for exactly that reason, because the conversion is easy to invert and the result is not obviously broken, just wrong.

Check the units before you change the numbers. Then change the numbers.

Three steps, and none of them require a view on where gold goes next.

First, write down what each distance setting was supposed to represent. If you cannot, that is the finding. Second, measure the instrument's recent typical range and check whether each setting still represents what it was meant to. Third, if you changed anything, test it on a demo across a losing stretch before it touches an account you paid for.

And be honest about which problem you are solving. If the settings drifted, recalibrate. If the strategy never had an edge, recalibrating a stop will not supply one, which is the diagnosis in is forex trading profitable.

Trading carries risk, and nothing here is a projection of where any instrument goes next.