Williams %R shows up on every platform, gets added in the first week of learning, and then quietly causes problems for years. Most traders use it wrong from day one, not because it is complicated, but because nobody explains what it is actually doing under the hood. I want to walk through it honestly, including the parts that will cost you money if you treat it like a buy and sell machine.
What Williams %R actually measures
The idea is simple. Over the last N bars (the default is 14), where is the current close sitting inside that range? If today's close is right at the top of the last 14-bar range, %R reads near 0. If it is right at the bottom, it reads near -100.
That is the whole thing. It is a position-within-range gauge. No complicated smoothing, no ratios. Just: of the total high-to-low distance over the lookback period, where did we close today?
The scale runs from 0 to -100. That negative scale trips people up constantly. A reading near 0 means price closed near the recent high. A reading near -100 means price closed near the recent low. The convention is backward from what feels natural, but once it clicks, it stays with you.
The standard zones are -20 and -80. Above -20 (close to 0) is considered overbought. Below -80 (close to -100) is considered oversold.
The mistake most traders make with it
They treat -80 as a buy signal and -20 as a sell signal. Every single time, automatically, like a reflex.
Here is why that gets expensive. Williams %R reaching -80 just means price closed near the bottom of its recent range. In a healthy downtrend, that is completely normal. Price is supposed to close near the low end of its recent range in a downtrend. Buying every -80 reading in a falling market is the same as buying every dip in something that is genuinely broken.
The indicator is not telling you the move is over. It is telling you where the close is sitting. Those are two completely different things.
The same trap exists on the other side. A %R reading above -20 in a strong uptrend is a sign of momentum, not a reason to sell. Fighting a market that keeps printing near its recent highs is a fast way to give money back.
The real blindspots (at least three worth knowing)
- It cannot tell a trend from a range. Williams %R behaves completely differently in those two environments, and it has no internal mechanism to distinguish them. In a range, the -80 and -20 zones are genuinely useful mean-reversion signals. In a trend, they are noise at best and traps at worst. You have to bring that context from outside the indicator.
- It is pinned by design at extremes. In a strong move, %R can sit at or near -100 for many consecutive bars. It is not stuck or broken. The market literally keeps closing near the low of the expanding range. Traders waiting for a bounce that the indicator supposedly promises end up holding losers while the market quietly keeps going.
- The lookback period changes the story completely. A 14-period %R on a 5-minute chart and a 14-period %R on a daily chart are measuring two entirely different things. The default setting is almost never the right one for your timeframe without checking it first. Most traders never adjust it and wonder why signals feel random.
- It is blind to volume and order flow. A close near the recent high on zero volume and a close near the recent high on massive buying volume produce the same %R reading. The indicator does not know which one it is looking at. That distinction matters enormously for whether the signal is worth acting on.
Which indicators complement it and why
- A trend filter, like a 50-period or 200-period moving average. This is the most important addition. It solves the biggest blindspot. When price is above a rising average, treat %R reaching -80 as a potential pullback entry, not a reversal. When price is below a falling average, treat -20 as a potential short entry, not a bottom. The trend filter turns an ambiguous signal into a directional one.
- ADX (Average Directional Index). ADX tells you how much of a trend is actually present. Low ADX means choppy or ranging conditions, which is where mean-reversion %R signals have some validity. High ADX means trending conditions, where you should be skeptical of any counter-trend %R read. ADX and %R are a natural pairing for exactly this reason.
- Volume or cumulative delta. Because %R is completely blind to who is participating in the move, a volume spike or delta shift confirming a %R extreme gives you something the indicator alone cannot. A -80 reading with real buying delta showing up underneath it is a different conversation than a -80 reading on thin air.
- Structure and key levels. Williams %R lining up with a known support or resistance is worth ten %R signals floating in empty space. The level gives you a reason for the move to stall there. %R alone does not.
How I use it as a tool, not a trigger
I do not take trades because %R hit -80. Period.
What I use it for is context within a plan that already has a direction. If I have a reason to be long (structure, trend, a key level holding), and I want to time the entry on a pullback, %R cooling down toward -80 in that context is useful information. It tells me the pullback has momentum behind it and we are near the low of the recent range, which is roughly where I want to be entering a long anyway.
The other thing I use it for is reading exhaustion. When %R has been pinned near 0 for an extended period and then starts rolling over, I take notice. Not as a trigger, but as a flag that says the upside pressure might be softening. Combined with delta or volume starting to fade, that flag carries more weight.
The way I think about it: %R tells you where the close is parked relative to recent history. That is genuinely useful information. But it knows nothing about why price is there, what the order flow looks like, or whether the trend is about to continue or turn. You have to supply all of that yourself. If you can do that, %R becomes a decent tool. If you cannot, it becomes a reliable way to lose money on both sides.
For more indicator breakdowns written the same way, head over to the full blog, where I cover setups, indicators, and real pattern testing.
Test your own version on 8 years of data
Here is a concrete starting point: take the rule "fade every %R -80 reading only when price is above the 50 EMA" and run it against 8 years of NQ data. Compare it to fading every -80 blindly. The gap between those two results will tell you more about context filtering than any article can.
That is exactly the kind of test the WFF Backtest Lab is built for.
Backtest Credits unlock the Qualified tier of the WFF Backtest Lab, where every strategy has already passed walk-forward testing on real futures data.
This is educational content, not financial advice. Nothing here is a recommendation to buy or sell any instrument. Past results do not guarantee future results. Most short-term traders lose money. Always test on your own data and trade your own plan.