The same range-position idea as stochastic, flipped onto a 0-to−100 scale — reading near zero means overbought, but it can stay there through a genuine trend.
Larry Williams introduced %R in 1973, measuring where the close sits in the recent high-low range, scaled 0 to −100.
Because %R runs upside-down relative to stochastic, traders new to it often misread which end is overbought.
Traders selling every "overbought" reading near zero learned it can stay there for weeks in a genuinely strong trend.
Serious use reads it alongside the trend, exactly like stochastic or RSI.
%R measures the close's position in the recent high-low range, but inverted — 0 means the close sat at the high, −100 at the low.
In a genuinely strong uptrend, closes keep landing near the top of the range, pinning %R near zero — the identical persistence problem seen throughout this course.
When price makes a new high but %R fails to reach a correspondingly extreme reading, that's a genuine warning worth investigating, exactly like RSI or stochastic divergence.
Through that stretch, %R stayed pinned near zero repeatedly, while price kept climbing well past where a threshold-seller would have exited.
Near that top, %R failed to make a correspondingly extreme reading on the final price high — a genuine divergence, confirmed once price itself rolled over.
%R reads −5. A trader new to the tool assumes this means "deeply oversold, buy." Correct?
Price is in a strong, sustained uptrend. %R reads −2 for the tenth straight day. A trader shorts immediately. Sound?
Price makes a new high, but %R fails to reach the extreme it hit on the prior high. What does this combination suggest?
Price above, %R below, watched tick by tick on the left — and the mark it leaves in the ledger on the right. A confirmed bearish divergence, its mirror image — and a false overbought that just kept climbing.
Price makes a new high. Judge whether %R genuinely fails to confirm — then call it: a real divergence, or just confirming momentum.
The classic error is misreading the inverted scale, or trading the raw extreme as a complete signal. The discipline is mechanical: internalize which end means overbought, then apply the same trend-and-divergence discipline as every other bounded oscillator.
Williams built a mirror of stochastic's own logic — proof that the specific math matters less than the discipline around it. Read the trend, demand real divergence — the scale is just packaging.
Markets are never wrong — opinions often are.