Standard RSI wisdom says overbought sits above 70 and oversold sits below 30. Watch RSI on gold during a genuine multi-week uptrend and that rule quietly breaks: the reading swings between roughly 40 and 80 for the entire run, never once touching 30, and every pullback to 40 turns out to be a buyable dip rather than a warning. The oscillator is not misbehaving. Its normal operating range has shifted with the trend, and that shift is information worth reading on its own.
The shift, defined
During a sustained uptrend, RSI commonly settles into what traders call a bull range, roughly 40 to 80, with pullbacks finding support nearer 40 rather than dropping all the way to 30. During a sustained downtrend, RSI commonly settles into a bear range, roughly 20 to 60, with rallies stalling nearer 60 rather than pushing to 70. The center of gravity for the whole oscillator moves with the trend, which is exactly why static 70 and 30 lines mislead so often once a real trend gets underway.
A worked example on gold
Gold trends from 2,380 to 2,460 over a month on the daily chart, using RSI set to 14 periods. Across that run, RSI dips to roughly 42, 44 and 41 on three separate pullbacks, never approaching 30, and each of those dips marks a spot where the uptrend resumed within a few sessions. Late in the run, RSI actually breaks below 40 and holds there through two full sessions for the first time in the entire move. That break is an early hint the bull range itself may be ending, worth far more attention than any single reading near 70.
Why this beats waiting for overbought and oversold
A trader fading every RSI reading above 70 during that same gold uptrend would have shorted directly into strength repeatedly and lost on nearly every attempt, since 70 and even 80 simply meant healthy trend, not exhaustion. Reading the range itself instead of the fixed thresholds keeps the oscillator answering a question suited to the actual regime rather than a generic rule that assumes every market behaves the same way. The same logic runs in reverse during a downtrend: a rally that pushes RSI up to 58 or 60 without breaking into the 60s and holding is still inside the bear range, and treating that bounce as a reason to buy ignores exactly the regime the reading is describing.
The catch
This read still lags, and it still needs several genuine touches of the current range before you can trust where its edges actually sit. One dip to 40 does not confirm a bull range on its own; three or four consistent touches across multiple pullbacks build that confidence. Declaring a shift from a single data point is just a new way of misreading the oscillator, with an extra step added. The regime read also says nothing about when the trend will actually end, only that its momentum character has shifted; treat a broken range as a reason to tighten management on existing positions, not as a standalone signal to reverse and trade the other direction.