Stochastic vs Williams %R
Stochastic measures where the latest close sits within the recent high-low range, counting up from the low and smoothing the result. Williams percent R measures that same position counting down from the high, so the two are one calculation displayed on opposite scales.
The honest version of this comparison is short: these are the same measurement. What differs is which end of the scale is zero, and whether there is any smoothing applied afterwards.
What each one is
The stochastic oscillator measures where the close sits within the recent high-low range, counting up from the low, and then smooths that result. Stochastic covers it.
Williams percent R measures the same position counting down from the high, with no smoothing at all. Williams percent R covers it.
The underlying question is identical. How far up its recent range did this bar close — both tools ask exactly that and nothing else.
Where they differ
Which end is zero. One reads high after a strong run; the other reads near zero. Same fact, opposite number, and it is the most common source of confusion between them.
Whether there is smoothing. The stochastic has smoothing settings, so its output can be steadied. Williams percent R has none, which makes it the rawer and noisier reading.
How much can be tuned. Three numbers on one and one lookback on the other, which means one of them can be adapted to an instrument and the other essentially cannot.
How well supported each is. The stochastic is on every platform and in every method description. Williams percent R exists everywhere and is discussed almost nowhere.
Where they agree
They are the same measurement. Plot both and the shapes are mirror images; there is no market condition in which one knows something the other does not.
Both pin in a trend. A sustained move holds either at an extreme for many bars, and treating that as exhaustion is the standard error with both.
Both fail in a range. When price is going nowhere it turns every few bars, and those short runs push the stochastic and Williams %R across any threshold constantly.
And both cost a round trip per signal acted on — the same fixed charge for either — which the unsmoothed one pays more often for the same idea. The stochastic’s signal line smooths away some of the crossings that Williams %R still reports.
Which one to use: Stochastic or Williams %R?
Run the stochastic. It has smoothing you can adjust, it is universally supported, and every rule you read elsewhere is written for its scale.
Run Williams percent R when you want the raw reading with nothing on top. No smoothing means no lag added, and it is the fastest version of this measurement available.
Run it if you genuinely prefer the inverted axis. That is a real preference, and it is honestly what this choice comes down to.
And never run both. The chart then carries one number twice, drawn upside down, and any agreement between the two is a mirror rather than a confirmation.
Why the inversion causes real mistakes
Because rules copy across looking correct. A threshold rule written for one scale reads perfectly sensibly on the other and does the opposite thing, and nothing on the chart flags it.
And because the familiar numbers appear in both. The same thresholds are quoted for each tool with opposite meanings, which is a trap laid by convention rather than by either indicator.
What the smoothing is actually worth
It sets the trade count. More smoothing means fewer threshold crossings, and the count is what decides whether a rule can pay for its own costs.
It absorbs single odd bars. A gap or a news spike can make one bar several times wider than the bars around it, and an unsmoothed reading goes to a limit on a bar like that. The stochastic’s smoothing spreads the effect over several bars instead.
It is the only adjustment available. Since the underlying question is fixed, smoothing is the whole of what you can tune, and the raw version leaves you nothing.
And it is where the fitting risk lives. Three interacting settings is three ways to shape the past, so pick them once and leave them rather than adjusting after every losing run.
The original data
The stochastic appears in 184 titles at a median of 11,915 across 136 channels, and Williams percent R in 35 at a median of 6,222 across 34.
184 videos on one and 35 on the other, across 136 and 34 channels. Five times the coverage for the smoothed version, and the rarer tool averages barely more than one video per channel — almost nobody returns to it, which says more about its adoption than any comparison of the formulas.
There is no conflict between them. They are the same reading on opposite scales — so what looks like two indicators disagreeing is one indicator drawn twice, and neither has told you anything the other did not.
How the videos on each are made
Measured on the same set of videos, stochastic runs to a median of 8.5 minutes a video and Williams %R to 6.9, leaving stochastic about 1.6 minutes longer. By title wording, 64% of stochastic videos are lessons (how to, tutorial, explained and the like), next to 51% of Williams %R videos. Stochastic therefore reads mostly as a lesson topic, while Williams %R is more often discussed, reviewed or argued about than taught step by step. 72% of the stochastic videos have not reached 50,000 views; the Williams %R figure is 83%. Exactly one title names both.
When it fails
The failure is running both and reading their mirror image as divergence, and the chart makes it look compelling. One line climbs while the other falls, which resembles exactly the pattern people are taught to watch for. It is not a pattern; it is what a mirror does. Trades taken on that basis are taken on an artifact of the display, and because the two will always move oppositely the signal is available every single day.
The second failure is copying a threshold rule across. It inverts entirely.
A third is treating a limit as exhaustion. It reports one-sided bars.
A fourth is expecting the raw version to be tunable. There is nothing to tune.
A fifth is optimizing three settings after losses. That is fitting.
And a sixth is adding a third bounded oscillator. It reads the same bars too.
Related
Stochastic covers the smoothed version counting up. Williams percent R covers the raw version counting down. And the relative strength index is a bounded oscillator that measures something else.
This is the clearest example on the site of two names for one thing. Where the close sits inside the recent range — that is both of them. One counts from the bottom and one from the top, and the argument about which is better is an argument about which way up you like your axis.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.