Concept
Williams %R
Williams %R is a Momentum & Oscillators concept. The Library holds 1 implementation — a working definition you can pull into Quant.
The standard Williams %R indicator
Williams %R exactly as classically defined — the faithful reference build of the original formula, free to run in Quant.
What is Williams %R?
Williams %R is Larry Williams' momentum oscillator measuring where the latest close sits inside the recent high-low range. It is computed as (highest high - close) divided by (highest high - lowest low) over a lookback window, 14 periods by convention, multiplied by -100. The scale runs from 0 (closing at the very top of the range) down to -100 (closing at the very bottom), with readings above -20 conventionally labeled overbought and readings below -80 oversold.
The oscillator is credited to Larry Williams, the futures trader and author, and dates from the early 1970s, spreading through his books and market writing. Early descriptions used a 10-day window; charting platforms later standardized on 14, the same default that took hold for RSI and the stochastic. The negative scale is a historical plotting convention rather than extra information, and some platforms display the axis as 0 to 100 with the sign dropped.
Mechanically it is the raw %K of the stochastic oscillator flipped onto a negative axis: %R equals %K minus 100. The practical difference is presentation and smoothing: %R is typically plotted raw with no signal line, so it reacts sooner than a slow stochastic and whipsaws more.
What a reading asserts is narrow and useful: it locates the latest close inside the recent high-low span, nothing more. That makes %R a fast, bounded expression of momentum in the range-location sense: closes migrating toward the top of the range are the signature of buying pressure, closes sagging toward the bottom the reverse. Because the scale is hard-bounded, %R pins to its extremes easily; a reading of -2 in a strong uptrend is normal, so the zones read best alongside trend context rather than as standalone triggers.
How to read Williams %R on a chart
%R plots in its own pane on a fixed 0 to -100 scale, so reading it is about zones and transitions rather than shapes.
- 1With the default 14-period lookback, mark the -20 and -80 reference lines: above -20 the close sits in the top fifth of the lookback range, overbought by convention; below -80, the bottom fifth, oversold.
- 2Establish the regime first: in a trending market expect %R to camp in one zone for long stretches, while in a range it oscillates between both.
- 3Watch transitions rather than levels: the hook back out of a zone, a failure to reach the far zone on a bounce, or oscillator extremes that no longer match price's extremes.
- 4Confirm against price structure or a slower gauge such as MACD before acting; %R alone is deliberately twitchy.
How it's calculated
Locates the latest close inside the highest-to-lowest range of the last n bars, on a 0 to -100 scale.
Developed by Larry Williams; the output is bounded between 0 and -100.
Readings above -20 are commonly treated as overbought and below -80 as oversold.
It is the Stochastic %K shifted: %R_t = %K_t - 100 for the same length.
How traders use it
- As an overbought/oversold gauge: extremes flag closes pinned to the edge of the range; in sideways markets these often precede fades, while in trends they can persist for many bars, so most workflows apply a trend filter before fading them.
- As a momentum-failure tell: in an uptrend, a rally that can no longer lift %R above -20 shows closes failing to reach the top of the range, the same weakening logic as a stochastic that stops reaching overbought.
- As a timing trigger inside a higher-timeframe trend filter: wait for %R to reach the oversold zone during an uptrend, then act on the hook back above -80 or a price trigger rather than the extreme reading alone.
- As a divergence tool: price making a new extreme while %R prints a shallower one sets up a regular bullish or bearish divergence read; the bounded scale compresses extremes, so many traders demand price confirmation before trading it.
- In multi-indicator combinations: library scripts pair %R with RSI and a moving average so the fast %R times entries that slower tools have already vetted, a division of labor that generalizes to other confirming gauges such as ROC.
Williams %R vs other momentum oscillators
Stochastic Oscillator: Raw %K and %R are the same number on different axes (%R = %K - 100). The stochastic convention adds smoothing and a %D signal line, so it fires later and cleaner; %R shows the unfiltered version.
RSI: RSI averages the magnitude of up closes against down closes, measuring the persistence of gains rather than position in a range. It moves more smoothly and pins to extremes less readily than %R, which reads only where the close sits in the recent range.
Stochastic RSI: Stochastic RSI applies the range-location formula to RSI values instead of price, an indicator of an indicator that reaches extremes even faster than %R. StochRSI is usually smoothed to stay usable.
CCI: CCI measures how far typical price has strayed from its own average in mean-deviation units and is unbounded, so it grades the size of an excursion. %R is hard-bounded and saturates at the extremes instead.
Concept family
Momentum & Oscillators
91 concepts mapped · 91 in the Library
Williams %R FAQ
Is Williams %R the same as the stochastic oscillator?
It carries the same information as the raw stochastic %K, mirrored: %R equals %K minus 100 and is plotted from 0 down to -100. The practical differences are that %R is conventionally unsmoothed and has no signal line, so it is twitchier than a fast stochastic. Signals translate directly: -20 and -80 in %R correspond to 80 and 20 in stochastic terms.
What counts as overbought and oversold on Williams %R?
Convention says readings above -20 are overbought and below -80 oversold, with some traders using -10 and -90 in strong trends. Treat them as statements about where price is closing within its recent range, not as reversal signals: in a persistent trend %R can hold the overbought zone for many bars while price keeps running.
What is the best setting for Williams %R?
There is no single best. The platform default of 14 is the common starting point, early descriptions used 10, and short-term traders sometimes go lower at the cost of more whipsaw. Longer windows make the extremes rarer and more meaningful. Test whatever you choose on the instrument and timeframe actually traded.
Can Williams %R stay overbought while price keeps rising?
Yes, routinely. Overbought means closes are printing near the top of the recent range, which is precisely what a strong uptrend does bar after bar. Persistent readings above -20 are as much a strength signal as a warning, which is why fading them without a trend filter performs poorly in trending conditions.
Does divergence work on Williams %R?
It is spotted the same way as on any oscillator: compare price extremes with %R extremes for regular divergence, or pullback extremes within a trend for hidden divergence. Because %R is unsmoothed and bounded it produces more candidates than RSI, many of them noise, so most practitioners require a price trigger before acting.
Turn Williams %R into a trading strategy.
Take the implementation from this page into Quant, then build on it, backtest it on real data, and keep refining it in conversation.