Williams %R: Momentum Oscillator Guide

What Williams %R Measures

Williams %R is a momentum oscillator created by Larry Williams in 1973. It answers a single question: where is the current close relative to the highest high and lowest low over the last N periods? Default lookback is 14 bars, and the output is a number between 0 and -100. The negative scale is the part that trips people up the first time they look at the indicator. The reading is always zero or below.

The formula is simple. Subtract the close from the highest high over the last 14 bars, divide by the range between the highest high and the lowest low over those same 14 bars, then multiply by -100. When the close is exactly at the highest high of the lookback window, the numerator is zero and the reading is 0. When the close is exactly at the lowest low, the numerator equals the range and the reading is -100.

Williams %R = (HighestHigh₁₄ − Close) / (HighestHigh₁₄ − LowestLow₁₄) × -100

Readings near 0 mean the close is at or near the top of the recent range. That's strong buying. Readings near -100 mean the close is at or near the bottom of the range. That's strong selling. A reading around -50 means the close is right in the middle of the range, which is the neutral midpoint.

Some charting platforms plot Williams %R on an inverted axis from 0 at the bottom to 100 at the top, so the line moves visually like the Relative Strength Index. The underlying number is still negative; only the display orientation changes. Always check which orientation your platform uses before reading levels off the chart.

The pane above plots Williams %R on AAPL under price. The line pushes toward its upper boundary during sustained advances and toward its lower boundary during selloffs, reaching those extremes faster and more often than smoothed oscillators do because it uses the raw close versus the range rather than the smoothed average gain ratio that the Relative Strength Index uses.

Williams %R vs Stochastic

The math behind Williams %R is almost identical to the stochastic oscillator's %K line. Both use the same lookback window. Both compare the close to the recent high-low range. The only differences are the orientation of the calculation and whether smoothing is applied.

The two formulas side by side

Stochastic %K = (Close − LowestLow) / (HighestHigh − LowestLow) × 100. The result is a positive number between 0 and 100. Close at the top of the range gives 100, close at the bottom gives 0.

Williams %R = (HighestHigh − Close) / (HighestHigh − LowestLow) × -100. Close at the top gives 0, close at the bottom gives -100.

If you plot stochastic %K and Williams %R on the same chart with the same period, the two lines are mirror images of each other. The peaks of one line up with the troughs of the other. They carry the same information. The choice between them is largely cosmetic preference.

Smoothing is the real difference

Williams %R is the raw oscillator. Whatever the close-to-range ratio is on the current bar, that's the number you see. No smoothing, no signal line, no second derivative.

Standard stochastic plots %K with a 3-period smoothing applied, then adds a %D line that's a further 3-period moving average of smoothed %K. The smoothing dampens the noise and adds lag. Williams %R skips both steps. That makes it faster to react and noisier in choppy conditions. The same data, processed less, with the predictable tradeoff.

Traders who want early warning of momentum shifts often prefer Williams %R because it flips into overbought or oversold territory before stochastic does. Traders who want fewer false signals usually prefer smoothed stochastic %D. Neither is objectively better. They sit at different points on the same speed-versus-reliability curve.

On NVDA, the high volatility means the Williams %R line hits its extremes constantly during fast moves. The pane above shows frequent touches at -20 and -80 with sharp reversals between them. A smoothed oscillator like the Relative Strength Index would register similar pushes but with more filtering, sacrificing the speed Williams %R is built for.

When to use Williams %R over stochastic

The honest answer is that they're substitutable in most setups. Pick the one your platform displays cleanly, the one that matches your own visual preference for negative or positive scales, and the one whose lag profile fits the timeframe you trade. Day traders on short bars often pick Williams %R for the extra responsiveness. Swing traders on daily charts often pick smoothed stochastic for the filtering. The choice rarely makes or breaks a strategy.

Trading the Extremes

The textbook reading of Williams %R says above -20 is overbought and below -80 is oversold. Beginners take that to mean sell above -20 and buy below -80. That rule fails the same way it fails on the Relative Strength Index, and for the same reason: extremes in a trending market confirm the trend rather than contradict it.

Extremes in trending markets

In a strong uptrend, Williams %R can sit between -20 and 0 for weeks. The close keeps printing near the top of the rolling range because that's what an uptrend does. A reading of -10 means buyers have been dominant. It does not mean sellers are about to take over. Shorting every push above -20 in a healthy uptrend is a quick way to give back capital.

The same logic runs in reverse during downtrends. Williams %R can stay pinned below -80 for an extended stretch when a stock is grinding lower and the close keeps landing in the bottom fifth of the range. Buying every dip below -80 in a downtrend is catching the falling knife the indicator is describing.

The cross-back-through is the signal

The actionable read on Williams %R extremes is the same as on the Relative Strength Index: wait for the cross back through the threshold before treating the reading as a signal. The indicator entering oversold tells you sellers were dominant. The indicator climbing back out of oversold tells you that dominance is fading.

A practical mean-reversion setup using Williams %R: the oscillator drops below -80 during a pullback in a broader uptrend, then crosses back above -80 from below. Enter on the cross. Stop just below the swing low that produced the oversold reading. The thesis is that selling pressure in the pullback has exhausted and buyers in the larger uptrend are stepping back in.

Adjusting thresholds to the asset

Williams %R thresholds are not universal constants. In strong trends the indicator never reaches the opposite extreme. A stock that's been trending up for months might never see Williams %R below -50, let alone -80. Waiting for a reading below -80 on that stock means waiting forever. The fix is to adapt the levels to the asset's recent behavior: if Williams %R consistently bounces off -50 during pullbacks in the uptrend, use -50 as your operative oversold level on that stock for as long as the trend is intact.

On MSFT during multi-year bull runs, the Williams %R line in the pane above tends to bounce off mid-range levels on pullbacks rather than ever reaching textbook oversold. In strong trends the lower extreme is rarely tagged, and the meaningful "oversold" inflection moves up the scale to wherever buyers actually keep stepping in.

Failure Swings

Failure swings are the pattern Larry Williams himself highlighted as the highest-quality signal the indicator produces. The structure relies on watching what happens after an extreme reading rather than on the extreme reading itself. The pattern works because it forces the oscillator to confirm that the previous trend has weakened before generating a signal.

Bullish failure swing

The full sequence is four steps. First, Williams %R drops below -80 into oversold. Second, it rallies back above -80. Third, on the next pullback in price, Williams %R declines but stops short of -80 — it makes a higher low than the oversold reading from step one. Fourth, the oscillator then breaks above the local peak it printed between the two declines.

The "failure" is the second dip's failure to return to oversold. Sellers pushed price down again but could not push the close as deep into the rolling range as before. Buyers are stepping in earlier each time. The break above the prior peak in Williams %R confirms that momentum has turned. This is a bullish structure that typically appears at the end of pullbacks within larger uptrends or at the bottoms of broader downtrends.

Bearish failure swing

The mirror sequence. Williams %R pushes above -20 into overbought. It pulls back below -20. On the next rally, the oscillator climbs but stops short of -20 — a lower high relative to the previous overbought reading. The oscillator then breaks below the local trough that separated the two rallies.

The failure here is the second rally's failure to reach overbought. Buyers pushed price up again but could not push the close back to the top of the rolling range. Sellers are showing up earlier. The break below the prior trough confirms momentum has rolled over. This is a bearish structure that often appears at the end of rallies within larger downtrends or at the tops of broader uptrends.

Why failure swings filter noise

A naive overbought or oversold reading is one data point. A failure swing is a sequence of four data points that collectively describe a momentum shift. The pattern requires the oscillator to first reach an extreme, then leave it, then attempt to return to it, then fail. That sequence takes time. By the time it completes, you have evidence that the previous side of the market is no longer in control.

Failure swings appear less frequently than threshold crosses but produce better risk-reward when they complete. The tradeoff is the same one that runs through all of technical analysis: filtering reduces the signal count and tightens the entries that survive the filter.

On QQQ, look for the four-step failure-swing structure on the Williams %R line above and you'll find spots where price reversed shortly after the pattern completed. Williams %R reaches the boundaries faster than the Relative Strength Index does, so on Williams %R the failure-swing extremes are usually at -80 and -20 rather than at 30 and 70.

Combining failure swings with price structure

The strongest version of a failure swing pairs the oscillator pattern with confirming price action. A bullish failure swing that completes while price is holding above a major support level is significantly higher quality than one that completes mid-air with no nearby reference. A bearish failure swing that completes as price rejects a major resistance level adds a second independent piece of evidence to the same trade.

Treat the failure swing as the entry trigger and the price structure as the setup that justifies looking for the trigger. The combination filters out the cases where the oscillator pattern fires but the broader chart offers no reason for the move to follow through.

Common Mistakes

Williams %R is simple to compute, which makes it easy to misuse. The mistakes below show up over and over in trading journals.

Treating extreme readings as automatic signals

The single biggest mistake is shorting because Williams %R crossed above -20 or buying because it crossed below -80. Extreme readings describe momentum, not entries. In a strong uptrend Williams %R can stay above -20 for weeks and the stock keeps grinding higher the entire time. In a strong downtrend the opposite. Use the cross back through the threshold as the actual trigger, not the entry into the extreme.

Ignoring trend context

Williams %R thresholds need to be interpreted against the larger trend. In an uptrend, -50 might be the operative oversold level because the indicator never reaches -80. In a downtrend, -50 might be the operative overbought level because the indicator never reaches -20. Mechanically applying the standard -20 and -80 levels regardless of trend produces signals that fire too rarely on one side and trap the trader on the other.

The simplest filter: check the slope of a moving average on price before trading any Williams %R signal. If the 50-day Simple Moving Average is rising, only act on bullish Williams %R setups. If it's falling, only act on bearish ones. This single rule eliminates most of the worst trades the indicator generates on its own.

Using Williams %R alone on choppy charts

Because it skips the smoothing that stochastic applies, Williams %R generates more whipsaws in range-bound markets than other oscillators do. The line ricochets between extremes when the stock is going nowhere, and every ricochet looks like a fresh signal. Before trading a Williams %R extreme, confirm the chart is actually trending. If it isn't, either widen your stops, switch to the smoothed stochastic, or skip the trade.

Reading the scale wrong

The negative scale catches new users. "Williams %R is high" is ambiguous because high could mean closer to 0 (overbought) or further into negative territory (oversold), depending on which way the reader is thinking about it. Be explicit when reading levels off the chart: -10 is near the top of the range, and -90 is near the bottom. The number being closer to zero means stronger recent buying.

Skipping the failure-swing structure

Most traders use Williams %R as a binary overbought/ oversold flag and never look at the four-step failure swing pattern. The threshold cross alone is the weakest signal the indicator produces. The failure swing was highlighted by Williams himself because it requires the oscillator to confirm the momentum shift across multiple bars. Skipping it and trading raw threshold crosses is using the simpler half of the toolkit.

Risen plots Williams %R alongside Relative Strength Index, stochastic, and price-condition alerts on every U.S. ticker chart, so traders can layer the oscillator into the rest of their setup without juggling separate tools.

Keep Learning