Stochastic RSI vs Williams %R: Which Turns First? (Real SPY Data)

Michael Whitman · Trading since 2017, full-time since 2020 · YouTube creator, 137K subscribers · Published 2026-09-09 · Updated 2026-11-02 · Reviewed 2026-11-01

Stochastic RSI ranks today’s RSI inside its last 14 readings and smooths it; Williams %R ranks today’s close inside the last 14 highs and lows, unsmoothed. At 111 SPY swing lows where both were oversold, Williams %R climbed back out first 78 times, the stochastic version 2.

Search for this pair and most results compare Williams %R with plain RSI or the plain stochastic instead. The stochastic RSI is a different animal: a range reading of an indicator, not of price. Below, both are worked by hand on one real SPY session, then timed against every confirmed SPY turn since 1996.

What each one is

Williams percent r places the close inside the recent high-low range. With the standard 14 bars it is −100 × (highest high − close) ÷ (highest high − lowest low), so it runs from −100 at the bottom of the range to 0 at the top. The −100 multiplier, which is how StockCharts’ ChartSchool writes Larry Williams' formula, turns the ratio so the line rises with price. Nothing is averaged. Williams %R covers it in full.

The stochastic relative strength index applies the same range idea to an indicator. First the 14-bar relative strength index is computed with Wilder’s smoothing. Tushar Chande and Stanley Kroll, who introduced it in their 1994 book The New Technical Trader, then took (today’s reading − lowest of the last 14 readings) ÷ (highest − lowest), a value from 0 to 1. Charting platforms multiply that by 100 and add two averages: %K, a 3-bar average of the raw value, and %D, a 3-bar average of %K. This page uses that platform form. Stochastic RSI covers the construction.

A worked example: SPY on 30 March 2026

SPY closed at $631.97 on Monday 30 Mar 2026, the lowest close of its March sell-off and a swing low in the test below. Here is each tool calculated from Yahoo Finance’s daily bars.

Williams %R. The highest high of the last 14 sessions was $680.08 (11 Mar) and the lowest low was $629.28, set that same Monday.

• Distance from the top: $680.08 − $631.97 = $48.11.
• Size of the range: $680.08 − $629.28 = $50.80.
• %R = −100 × 48.11 ÷ 50.80 = −94.70, deep in oversold (−80 or below).

Stochastic RSI. The 14-bar RSI closed at 27.30, which was itself the lowest of its last 14 readings; the highest was 42.75 (11 Mar).

• Raw stochastic RSI = (27.30 − 27.30) ÷ (42.75 − 27.30) × 100 = 0.
• The raw value on the three latest sessions was 20.08, 0 and 0, so %K = (20.08 + 0 + 0) ÷ 3 = 6.69.
• %K on those sessions was 39.32, 27.33 and 6.69, so %D = 24.45.

What the arithmetic shows. On Friday 27 Mar the raw stochastic RSI was already at 0 and %R was −98.05, yet %K read 27.33, outside its oversold line, because the average still carried the 61.91 and 20.08 of the two sessions before. Williams %R had been at −80 or lower since Thursday 26 Mar; %K only got there on the low itself. The next day, 31 Mar, SPY closed at $650.34: %R jumped to −53.37 and %K to 33.33, so both left oversold together. Then the raw stochastic RSI sat at exactly 100 for seven straight sessions.

Where they differ

The input. Williams uses the bar’s high and low, so a long wick shows up. The RSI is built from closes only, so the stochastic version never sees intraday range at all.

What the range is measured against. Williams compares the close to real price extremes. The stochastic version compares an RSI reading to other RSI readings, a narrow band that can sit close together even in a week when price moved a lot.

Smoothing. Williams shows every jump. %K and %D average the raw value twice, which makes a cross easy to name in a rule and adds a lag of a session or two, the lag that shows up in the timing data.

How often each sits on its edge. Any new 14-reading high or low in the RSI puts the raw stochastic version at exactly 100 or 0. Williams reaches exactly 0 only when the close equals the highest high of all 14 bars.

The original data

Which one turns first: the test. Yahoo Finance SPY daily bars, high, low and close as traded, from 2 January 1996 to 30 October 2026, 7,759 sessions (earlier bars only start the RSI). A swing low is a close lower than each of the 10 closes before it and the 10 after it, and at least 3% below the highest close of the previous 20 sessions; a swing high mirrors it. That found 164 lows and 173 highs.

For each turn, the test recorded whether each tool was in its zone (−80 or below for Williams, %K 20 or below; −20 and 80 at highs) at the turn’s close, when that stay in the zone began, and on which session the tool closed back outside it.

Getting out: Williams %R, almost every time. At the 111 SPY lows where both were oversold on the low day, Williams %R closed back above −80 first 78 times, the stochastic version first 2 times, and they left on the same session 31 times. At the 117 highs where both were overbought, Williams left first 56 times to 12, with 49 ties. The stochastic %K left a median 2 sessions after the low, Williams 1.

Getting in: it depends on the side. At the lows, %K’s oversold stay began first 65 times against 25 for Williams, with 21 on the same session (sign test, p < 0.001). At the highs it ran the other way: Williams turned overbought first 58 times against 35 (p = 0.022). The 30 March example above was one of the 25 lows where Williams got there first.

On the turn day itself, Williams was in its zone more often: at 143 of 164 lows (87.20%) against 123 (75.00%) for %K, and at 160 of 173 highs (92.49%) against 120 (69.36%). The stochastic version’s own lowest reading within 10 sessions either side came before the price low at 91 of the 164 lows, which fits the RSI bottoming out before price does.

The exit result held everywhere; the entry result is weaker. On IWM, a small-company fund, %K’s earlier entry at lows sits within chance, and at IWM highs Williams entered first 58 times to 41 (p = 0.11). Nasdaq’s SPY bars found the same 91 turns from January 2017 as Yahoo’s and the same zone readings at every one. Every turn, with both readings, is in the turns file.

What this does not show. The turns were picked with hindsight, 10 sessions after the fact. The test times the two tools around turns that happened; it says nothing about how often a zone reading is followed by one.

And on that, the earlier test is plain. For each tool, a zone entry (first close inside after a close outside, entries at least 10 sessions apart) was checked against SPY’s close 10 sessions later, dividends included, from 1996 to 2 Oct 2026. Any session was followed by a higher close 61.43% of the time.

All four are within chance. Every entry is in the zone entries file, and every daily value of both tools in the daily readings file.

Pinned at the edge. Over that 1996 to 2 Oct 2026 sample, the raw stochastic version sat at exactly 0 or 100 on 29.53% of sessions; Williams percent r sat at exactly 0 or −100 on 1.11%. Their daily changes correlated 0.50 on SPY and 0.49 on IWM: related, but two different readings.

Which one to use: Stochastic RSI or Williams %R?

Run williams percent r when the question is whether a move has ended. It uses the real high and low, it is in its zone on the turn day more often, and it climbs out of the zone before the stochastic version at almost every turn in the data.

Run the stochastic relative strength index when a rule needs a discrete event. A %K and %D cross is easy to write down and test. Accept that the smoothing costs a session or two on the way out, and that the raw value spends more than a quarter of all sessions pinned at 0 or 100.

Run neither as a trigger against the trend. Zone entries on both were followed by ordinary results on SPY, and on IWM both tools turning oversold together was followed by a higher close 10 sessions later 55.71% of the time, against 56.88% for any session (IWM agreement file).

When it fails

Reading agreement as confirmation. Both respond to the same falling closes, so two tools in oversold is closer to one observation than two. The useful days are the ones where they disagree, like 27 Mar 2026, and those say something about the inputs, not about direction.

Treating a fast exit as a buy signal. Williams leaving oversold first is a description of the bounce that already happened. In a falling market it leaves and re-enters again and again.

Treating a pinned 100 as extra strength. After the March 2026 low, the raw stochastic RSI sat at exactly 100 for seven sessions in a row. A reading that common carries little information by itself.

Tuning the lookback until the last turn is caught. That fits one sample and nothing beyond it.

Ignoring costs. Either tool enters a zone dozens of times a year on a daily chart, and each one acted on is a round trip paid before anything is known.

Questions people ask

Which is faster, Stochastic RSI or Williams %R?

Out of a zone, Williams %R: after 78 of 111 SPY lows where both were oversold it left first, against 2 for Stochastic RSI’s %K. Into a zone it depends on the side: %K got oversold first at SPY lows, %R overbought first at SPY highs.

Is Stochastic RSI the same as Williams %R?

No. Williams %R is the fast stochastic of price, read upside down; Stochastic RSI is a stochastic of the RSI, then smoothed. Their daily changes correlated 0.50 on SPY.

Which indicator works best with Stochastic RSI?

Nothing in this data makes a pairing work as a signal. A price-based reading such as Williams %R adds what the stochastic RSI cannot see, the high and low, but on IWM both turning oversold together was followed by ordinary results.

Which is better, RSI or Stochastic RSI?

They answer different questions: RSI is the momentum reading, Stochastic RSI ranks that reading against its own recent range. The RSI vs Stochastic RSI comparison covers that pair.

Related

Williams %R covers the raw range-position reading and why it matches the fast stochastic. Stochastic RSI covers the twice-derived construction and the signal line. And RSI covers the measure the stochastic version is built from.

This page is educational, not financial advice. Test every idea on your own charts before risking money.