Relative Volume (RVOL): What 40,212 Trading Days Show

Michael Whitman · Trading since 2017, full-time since 2020 · Published 2026-09-25

Raw volume numbers are hard to read on their own. Fifty million shares is a quiet day for one ticker and a frenzy for another. Relative volume fixes that by dividing each day by its own recent normal, which is why many day-trading scanners show it as a column. This page sets out the formula, works one real example, and measures what high and low readings were followed by on six US symbols.

How it forms

The calculation is a single division. Take the volume of the current session and divide it by the average volume of a lookback window. A reading of 1.0 means an ordinary day. A reading of 3.0 means three times the usual number of shares changed hands.

Two choices sit inside that division. The first is the length of the lookback. Tools differ here, and a longer window reacts more slowly to a change in regime. The second is whether today is included in the average. Leaving it out keeps a spike from diluting its own baseline, so this page leaves it out.

Intraday versions compare like with like. During the session a scanner cannot know the full-day total yet, so a common intraday version compares the volume traded so far with the average traded by the same time of day. At 10:00 a.m. that is a small, noisy sample. The daily figure used below is the settled version, calculated after the close.

Unusual volume has a reason behind it. Earnings, a guidance change, an index addition, a macro release, a large holder selling, or options and futures expiring can all pull extra shares into the market. RVOL flags that something happened; it does not say what, and it does not say which side won.

The definition used here

RVOL20 is the day’s volume divided by the mean volume of the 20 sessions before it. The data are daily bars from the Yahoo Finance chart API for SPY and five liquid large caps, Apple (AAPL), Microsoft (MSFT), JPMorgan Chase (JPM), Exxon Mobil (XOM) and Coca-Cola (KO), downloaded on 25 September 2026. Volume is split-adjusted by the source.

The sample runs from 1 February 2000 to 24 September 2026. Every day needs 20 prior sessions that fall in 2000 or later and one session after it, which leaves 6,702 days per symbol and 40,212 in all.

Each day goes into one of five buckets: under 0.7, 0.7 to 1.0, 1.0 to 1.5, 1.5 to 2.0, and 2.0 and up. For each bucket the page reports the median absolute close-to-close move on the day itself, the median absolute move on the following day, and how often that following day closed higher.

A worked example

On Friday 18 September 2026, 65,395,100 SPY shares traded. The 20 sessions before it, 20 August to 17 September, added up to 796,573,800 shares. Divide by 20 and the average is 39,828,690 shares.

RVOL is 65,395,100 divided by 39,828,690, which is 1.64. That was the busiest SPY session in the window above, about two-thirds more shares than normal.

Yet price barely moved. SPY closed at $761.69, down 0.12% from the prior close of $762.60. The calendar offers the likely reason: 18 September was the third Friday of the month, which Cboe’s 2026 expiration calendar lists as the standard expiration day for equity, index and ETF options. Plenty of shares changed hands on a flat day.

Compare 5 June 2026. SPY traded 93,989,400 shares against a prior 20-session average of 46,822,605, an RVOL of 2.01, and closed down 2.58%. The two readings are not far apart, but one came with a flat close and the other with SPY’s largest RVOL of the year so far. The ratio alone cannot tell them apart.

The original data

Across the 40,212 symbol-days, busy days were big days. On days with RVOL under 0.7, the median absolute move that day was 0.49%. In the 1.0 to 1.5 bucket it was 1.01%. At 2.0 and up it was 2.52%, more than five times the quiet-day figure.

The buckets are very uneven in size. 6,073 days fell under 0.7, 17,789 between 0.7 and 1.0, 12,859 between 1.0 and 1.5, 2,452 between 1.5 and 2.0 and 1,039 at 2.0 or more. The three middle buckets, in order, showed same-day medians of 0.66%, 1.01% and 1.64%, next-day medians of 0.74%, 0.81% and 1.03%, and next days that closed higher 51.6%, 51.9% and 51.4% of the time. Pooled, all 40,212 days had a median move of 0.78% on the day and the same 0.78% on the day after.

The next session carried some of that size forward. Its median absolute move rose from 0.73% after the quietest days to 1.15% after days at 2.0 or higher. The share of next days that moved 2% or more, up or down, went from 978 of 6,073 (16.1%) in the lowest bucket to 322 of 1,039 (31.0%) in the highest.

Direction stayed close to a coin toss. The next day closed higher 51.5% of the time in the lowest bucket, 51.9% in the middle one and 50.4% in the highest, against 51.7% for all 40,212 days. High relative volume came before bigger moves in both directions, not before rallies.

Some of that is the regime, not the day. Busy sessions tend to arrive in turbulent stretches, when every day is moving more. To check, each next-day move was divided by the average absolute move of the 20 sessions ending on the signal day. The median of that ratio was 0.68 after RVOL under 0.7 and 1.07 after RVOL of 2.0 and up, against 0.80 for all days. So the effect survives the adjustment, though the busiest bucket’s lead over all days narrows, from 47% above the all-day median on the raw measure to 34% on the adjusted one. The bucket summary is in the RVOL bucket file.

Calendar days that inflate it

Quarterly expiration Fridays are the clearest case. Of SPY’s 6,702 measured days, 105 fell on the third Friday of March, June, September or December, the standard options expiration day in a quarter month. Two such Fridays, 21 March 2008 and 19 June 2026, were market holidays with no session in the data, so they are not counted. The median RVOL of the 105 was 1.23, against 0.95 for every SPY day; 30 of them reached 1.5 or more and 7 reached 2.0 or more.

Their price moves were ordinary. The median absolute SPY move on those 105 days was 0.61%, close to the 0.54% median for all SPY days in the sample. That is the pattern 18 September 2026 showed: extra shares, little new information.

Readings of 2.0 or more are uncommon on large, liquid names. They showed up on 2.2% of SPY days, 3.4% for AAPL, 2.8% for MSFT, 2.9% for JPM, 1.6% for XOM and 2.6% for KO. A screen that returns dozens of 2.0+ names every morning is scanning a far wider universe than these six, and on thinner stocks one order can move the ratio.

When it fails

Reading direction into it is the first mistake. A red candle on RVOL of 3 is often called “distribution” and a green one “accumulation.” On these six symbols, the day after a 2.0+ reading closed higher about as often as any other day, and the candle’s color did not rescue it: after 453 green 2.0+ days the next day closed higher 47.0% of the time, and after 586 red ones 53.1%, both within a few points of the 51.7% for all days and the opposite way round to the labels.

A thin baseline makes the ratio jumpy. The 20 sessions after a holiday stretch, or after a quiet August, pull the average down, and a normal day then reads as unusual. Checking the raw share count next to the ratio catches this.

Intraday RVOL early in the session is fragile. A single block print in the first ten minutes can push the reading to, say, 5.0 at 9:40 and leave it near 1.2 by lunch. Traders who act on the opening reading are acting on a small sample.

Comparing tickers by RVOL alone mixes very different markets. A 2.0 reading on SPY involves tens of millions of extra shares; on a small cap it may be a few hundred thousand. Liquidity, spread and slippage do not scale with the ratio.

And calendar days pass as news days. Expiration Fridays lift volume without saying anything about the business, as the September 2026 example and the 105 quarterly expiration Fridays above show.

Related

The volume page covers what the raw number counts and why it is the one chart input not built from price. Volume analysis shows how the choice of baseline changes what looks like a high-volume bar, which is the same problem RVOL’s lookback has to settle. And a stock screener is where most traders meet RVOL first, usually as a filter sitting next to price and gap size.

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