Williams %R

Add Williams %R to a chart and the first thing that jars is the scale. Every other oscillator you have met runs from 0 to 100. This one runs from −100 at the bottom to 0 at the top — all negative numbers, with zero as the ceiling. A reading of −8 sits near the top; a reading of −92 sits near the bottom.

That single design choice trips up almost every beginner, because it inverts the plain-arithmetic instinct. Your gut says −8 is "very low" and −92 is "higher." The scale says the exact opposite: −8 is the strong end and −92 is the weak end. Get that backwards — and people constantly do — and you will read every signal upside-down.

Here is the reassuring part, and the whole point of this short module. Once you understand the scale, there is almost nothing new to learn, because Williams %R is the Stochastic you already know, plotted on a flipped axis. Same idea, same strengths, same trend trap — just measured from the top down.

Williams %R, created by the trader Larry Williams, asks the identical question the Stochastic asks: where did the close land inside the recent high-low range? The only difference is the reference point. The Stochastic measures the close's distance up from the range low. Williams %R measures the close's distance down from the range high.

So when a stock closes right at the top of its recent range — maximum strength — its distance below the high is zero, and Williams %R reads 0. When it closes at the very bottom — maximum weakness — its distance below the high is the full range, and %R reads −100. Everything in between scales linearly. In fact the relationship to the Stochastic %K is exact: Williams %R is essentially −(100 − %K). It is the same measurement, mirrored.

Why does a mirror image of an existing tool exist at all? Partly history — it predates some of the smoothed Stochastic variants — and partly taste; some traders simply prefer reading strength from the top down, with the zero line as the "close at the highs" reference. But you should hold one fact firmly from the start: because it is the Stochastic flipped, putting both on a chart gives you no extra confirmation. Two upside-down views of the same numbers cannot vote independently. It is one tool wearing two costumes.

Two bands are drawn, and their positions are the mirror of the Stochastic's:

• −20 — the overbought line, near the top (near zero). Above it (i.e. between −20 and 0), the stock keeps closing near the top of its range.
• −80 — the oversold line, near the bottom. Below it (between −80 and −100), the stock keeps closing near the bottom.

The midline at −50 is the balance point. The mental translation you must automate: closer to 0 = stronger; closer to −100 = weaker. Say it until it stops feeling backwards.

Once the scale clicks, the reading is pure Stochastic. In a range it cycles between the bands and does a decent job flagging the edges. The crosses of the −20 and −80 lines, taken with the price context, are its most usable signals.

And it inherits the Stochastic's flaw exactly. In a strong uptrend, Williams %R glues itself near the zero line and stays there while price runs — the mirror of the Stochastic pinning above 80. Every day it prints "overbought" near zero, and every day the trend ignores it. The inverted scale does not change the physics one bit.

Walk one composite chart. illustrative

The stock below is range-bound between roughly ₹97 and ₹106 — the same friendly territory the Stochastic likes. Underneath, picture Williams %R on its inverted scale.

At the floor, the stock is closing near the bottom of its range, so %R sinks toward −92 — the weak end (the up-arrow marks the point where buyers step back in). At the ceiling, it is closing near the top, so %R lifts toward zero — the strong end (the down-arrow). Read the minus signs the right way round and the rhythm is identical to the Stochastic's: strong at the top, weak at the bottom, cycling in a range.

The discipline is identical too. Confirm from the price that you are in a range before you trust the band touches, and remember that the instant the range breaks into a trend, %R will pin at one end and its band signals will turn to noise. This is not a new tool with new rules — it is the same tool, and it fails in the same place. Re-skinning an oscillator with a different scale does not give it a new edge; the limits travel with the maths.

Because it is the Stochastic in a mirror, every limit from that module applies without change. It cannot tell a range from a trend — the fact that decides how to read it is one it cannot supply. It cannot see the company behind the price; a run of weak closes looks the same whether it is a routine dip or a business unravelling. %R maps where the close sat in the range, never why.

It also cannot give you independent confirmation of anything the Stochastic already told you, because it is the Stochastic. Traders sometimes stack Williams %R, the Stochastic, and the Stochastic RSI together and feel reassured when they line up — but they are near-copies of one measurement, so their agreement is guaranteed, not informative. Genuine confirmation comes from tools that look at different things — volume, trend structure, a moving average — not from three views of the same close-in-range calculation. Three mirrors of one reading do not multiply the odds; they just repeat them.

And it will not survive curve-fitting any better than its siblings. Tune the lookback and the −20/−80 levels until they nail an old chart, and you have fitted the tool to that chart's accidents. An oscillator tuned to make the past look easy has been fitted to noise.

1. Reading the scale backwards. The headline error unique to this tool: treating a number near zero as "low" and a number near −100 as "high". Closer to zero is stronger. Burn it in.

2. Trading the bands in a trend. Shorting near-zero readings in an uptrend, or buying near-−100 readings in a downtrend, loses steadily while the trend runs — the same failure the Stochastic has.

3. Double-counting it with the Stochastic. Putting both on the chart and calling their agreement "confirmation." They are the same measurement flipped; they cannot confirm each other.

4. Skipping the range-or-trend check. As always, the price context decides the meaning, and %R cannot tell you the context.

5. Curve-fitting the settings. Sliding the lookback and levels until the past chart glows green — the backtest trap, once more.