OptiNod Academy

Williams %R — Why an Indicator Based on Close Position Stays Pinned in Trending Markets

Williams %R quickly shows where the close sits within the recent high-low range, but the -20/-80 thresholds only matter in range-bound markets.

Williams %R was formalized by Larry Williams in the 1970s, and its calculation fits on one line. Subtract the current close from the highest high over the last N bars, divide that by the difference between the highest high and lowest low over the same period, then multiply by -100. If the close is near the N-bar high, the value is close to 0. If it is near the N-bar low, the value is close to -100. The calculation is almost the same as Stochastic %K, except the sign is inverted and the scale runs from 0 to -100. %R tells you the close's relative position inside the previous N-bar range.

The popular rule is simple: above -20 is overbought, below -80 is oversold. Williams himself presented those levels in his book. The problem is that if you stop at that one line, %R gets treated like a cousin of RSI. Traders sell above -20, buy below -80, and use the extremes as a mean-reversion tool. In a trending market, that rule leads to mounting losses.

In a strong uptrend, each bar tends to close near the high of the previous N-bar range. By the %R formula, when the close is pinned near the high, the value stays close to 0, so %R can remain above -20 for days. It gets stuck in the so-called overbought zone. If you read that condition as a sell signal, you are shorting into the middle of the trend. Traders who use Williams %R more carefully first ask how long it has stayed above -20, then decide whether the market is ranging or trending before interpreting the threshold.

Williams %R Measures Where the Close Lands Inside the Recent Range

Read the formula directly and the indicator's purpose becomes clear: (highest high - close) / (highest high - lowest low) x -100. The numerator shows how far the close is below the N-bar high. The denominator shows the full width of the N-bar range. If the close equals the N-bar high, the numerator is 0 and %R is 0. If the close equals the N-bar low, the numerator equals the denominator and %R is -100. %R reports one thing: where the close finished inside the recent range, regardless of whether price is expensive or cheap.

That distinction matters in live trading. While BTC rose from $75,572 on November 6, 2024 to $97,900 on November 24 in 18 bars, the daily close finished near the top of the prior 14-bar range almost every day. Price became more expensive each day, but the close's relative position remained pinned to the top of the range. The 14-period %R was above -20 on 18 of 19 bars in that stretch. For the entire 32% rally, %R stayed in the overbought zone.

Once you accept that %R measures position, a persistent reading above -20 reads as clear evidence of a strong trend. When each bar closes near the top of the range, buyers are overwhelming that day's selling pressure again and again. This is the first decision point in interpreting Williams %R.

The -20/-80 Thresholds Matter Only in Ranges

Williams %R works as a mean-reversion tool only in range-bound markets. In a range, price moves between the upper and lower boundaries. If the close finishes near the top of the range (%R above -20), price is likely near range resistance and has room to pull back. If the close finishes near the bottom (%R below -80), price is likely near range support and has room to rebound. Under those conditions, the thresholds can support an entry.

BTC's sideways move from July to early September 2024, between $54,000 and $70,000 for more than two months, is a textbook case. Across the 67 bars in that range, the 14-period %R moved above -20 fifteen times and below -80 nine times. On July 15, the close reached $64,724 near the top of the range and %R climbed to -1.54 before price pulled back within a few days. On August 5, the close reached $54,018 near the bottom of the range and %R fell to -76 before price rebounded. The same thresholds worked at both ends of the range.

Use the same rule in a trend and one threshold is rarely touched while the other is hit repeatedly. In an uptrend, %R often holds above -20 and rarely reaches -80, so a rule that sells above -20 keeps producing sell signals in the middle of the trend. That is why you must first decide whether the market is ranging or trending before applying the thresholds mechanically.

First Classify the Market by the Percentage of Bars Pinned Above -20

There is a simple way to separate ranges from trends with one number: look at the percentage of bars over the previous 30 to 40 bars where %R was above -20. That percentage describes the market state.

• Below 10%: Range-bound market. Occasional touches of -20 can become range-top sell candidates.
• 10-30%: Weak trend or late-stage range. Mean-reversion setups start to lose reliability.
• Above 30%: Strong uptrend. Turn off the -20 sell rule. It is normal for %R to stay above -20, and a clear move back below -20 becomes the informative event.

BTC's October 14 to November 24, 2024 stretch shows how this classification works. During those 42 bars, %R was above -20 on 71% of bars, while price rose from $66,084 to $97,900. Since the percentage was far above 30%, the -20 sell rule should have been disabled from the start. Anyone using that sell rule during this period had to buy back at higher prices after every short.

Apply the same analysis to the percentage of bars below -80 to identify downtrends. If both percentages are below 10%, the market is ranging. If one side is above 30%, the market is trending in that direction. Classify the market first, then interpret the thresholds.

In Trends, a Persistent Reading Above -20 Confirms Strength

If you flip a persistent reading above -20 from a sell signal into bullish confirmation, Williams %R becomes useful in trending markets. When each bar closes near the top of the N-bar range and %R keeps holding above -20, it is a clear sign that buying pressure is repeatedly overwhelming that day's selling. For trend followers, that persistence confirms the trend is still intact.

The informative event is the clean break back below -20. While BTC kept %R above -20 for 18 bars from November 6 to November 24, 2024, price rose without any meaningful break. Then on November 25, BTC closed down at $93,010 and %R dropped straight to -45. After several days of holding above -20, the close finished below the middle of the range for the first time. From that point, price moved into several days of sideways action. The release from the -20 pin was the first signal of a pause in the trend.

Using Williams %R for trend following is simple. During a period pinned above -20, keep holding. If %R closes below -50, treat it as a sign the trend is taking a pause. The key is to turn off the -20 sell rule in trending markets and switch to this interpretation instead.

It Is Almost the Same as %K, but Noisier Without Smoothing

Williams %R and Stochastic %K measure almost the same information in almost the same way. Stochastic %K uses (close - N-bar low) / (N-bar high - N-bar low) x 100 on a 0-100 scale. %R looks at the relationship between the close and the high in the numerator and uses an inverted 0 to -100 scale. Add 100 to %R on the same bar and it will almost match %K. Both indicators measure the close's position inside the range.

The decisive difference is smoothing. Standard Stochastic smooths the raw %K once with an SMA, then applies %D as a second smoothing layer to filter noise. Basic Williams %R shows the raw, unsmoothed value. Without smoothing, you get faster response, but you also get more false signals on short timeframes. On 5-minute or 15-minute charts, unsmoothed %R can cross -20/-80 several times in an hour, and most of those crosses are noise.

If you use Williams %R without smoothing, it is more sensible to use it on longer timeframes or lengthen the lookback period. On volatile BTC daily charts, a 21-period setting reduces noise compared with 14 periods. On shorter timeframes, you can also add SMA(3) directly on top of %R as a smoothing step. If you ignore the lack of smoothing and use a 14-period %R unchanged on short timeframes, you will get pulled into repeated false crosses.

Divergence Shows Trend Fatigue Before the Trend Breaks

Williams %R divergence appears when price makes a new high but %R fails to exceed its prior high. That means the close is making a new price high, but its relative position inside the previous N-bar range is not as strong as it was at the prior high. The buying pressure that had been pushing closes to the top of the range is fading. Price is still higher, but the close's position within the range is weakening.

When BTC made its then-high at $108,353 on December 17, 2024, the 14-period %R was -12. At the previous November 22 high of $99,588, %R had reached -2.92. Price made a higher high, but the close's relative position was not as strong. Then on December 19, the close fell to $97,461 and %R dropped straight to -76. Price corrected into the $92,000 area by year-end. This was a case where divergence showed trend fatigue before the price correction.

Still, divergence works as a warning of trend fatigue. In trending markets, divergence often appears two or three times while price keeps rising. When divergence appears, check whether price structure is breaking too. It becomes an entry area only when paired with confirmation such as a close below the prior swing low.

Range Mean-Reversion Setup

A Williams %R mean-reversion setup starts with confirmation that the market is range-bound. Only after classifying the market should you use the thresholds as entry evidence.

• [ ] Market state: BTC daily has been in a clear horizontal range for more than two months, and over the previous 30 bars, both the percentage of bars above -20 and the percentage below -80 are under 30%.
• [ ] Entry condition: Price is near the bottom of the range, and the 14-period %R moves below -80, then recovers above -80 on a closing basis.
• [ ] Entry: Buy at the close of the bar where %R recovers above -80.
• [ ] Stop: Place it below the prior swing low under the range bottom.
• [ ] Target: Near the top of the range, where %R reaches -20.
• [ ] Invalidation: If the close clearly breaks below the range bottom, treat the range as broken and exit.

This setup works only in ranges. If you enter on a recovery above -80 without checking the previous 30-bar %R persistence percentages, you may be trying to catch a falling market in the middle of a downtrend. Confirm the market state before evaluating the entry condition.

Three Common Ways Traders Misuse Williams %R

Selling above -20 and buying below -80 in trending markets. This is the most common loss pattern. In a trend, only one threshold is hit repeatedly, so you keep entering against the trend. Before using the thresholds, check the percentage of bars above -20 over the previous 30 bars. If it is 30% or higher, turn off the -20 sell rule.

Using unsmoothed %R unchanged on short timeframes. Basic %R has no smoothing, so 5-minute and 15-minute charts produce many false crosses. On short timeframes, lengthen the period to 21 or add SMA(3) on top of %R as a smoothing step.

Entering on divergence alone. In trending markets, %R divergence often repeats several times while price keeps rising. Divergence is a warning of trend fatigue. It needs confirmation from price structure, such as a close below the prior swing low, before it becomes an entry area.

Two Higher-Level Checks for Williams %R

A Williams %R setup becomes more reliable when two things line up.

First is the market state on the higher timeframe. A 1-hour range mean-reversion setup loses reliability if it conflicts with the 4-hour trend. If the 4-hour chart is in a strong uptrend and you take a -20 sell on the 1-hour chart, you are entering against the higher-timeframe trend. Before taking a 1-hour setup, check the 4-hour %R percentage above -20.

Second is price structure. A buy on recovery above -80 matters only when price is actually near the bottom of the range. If price is in the middle of the range and only %R has dipped below -80, the stop distance to the range bottom widens and the mean-reversion case weakens. Williams %R is a fast tool for showing the close's relative position. That position becomes meaningful only after you classify the market state and read the price structure. This is why trades based on a single threshold so often lose money in trending markets.