NR7: The Narrowest Day in Seven, Tested on 1,272 SPY Days

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

Markets alternate between quiet stretches and busy ones, and the NR7 is one of the simplest ways to spot the quiet end. It asks a single question of each day: was this range smaller than every one of the six before it? This page defines it exactly, walks through two NR7 days from September 2026, and tests what followed on SPY and five large caps.

How it forms

The range is the only input. High minus low, with no reference to the open, the close or volume. If today’s range is smaller than each of the previous six, today is an NR7, short for “narrow range seven.”

The idea comes from short-term pattern research. NR7 comes from Toby Crabel’s 1990 book, Day Trading with Short Term Price Patterns and Opening Range Breakout, and Crabel’s methods are among those covered in Street Smarts (1995) by Linda Raschke and Laurence Connors. The reasoning is that volatility moves in cycles, so an unusually tight day hints that a wider one is coming.

The trade built on it is a two-sided breakout. A trader places a buy order just above the NR7 high and a sell order just below the NR7 low, and whichever fills first sets the direction. The other order is then canceled or turned into a stop.

Relatives use other lookbacks. NR4 compares with the prior three days, and some traders combine NR7 with an inside bar, a day whose high and low both sit within the previous day’s, to demand an even tighter setup.

The definition used here

An NR7 day has a high-minus-low range strictly smaller than each of the six previous sessions' ranges. A tie does not count. SPY was quoted in fractions of a dollar until 2001, when equal ranges were common, and those ties are kept as ties: rounding the old prices to decimals would have added 16 false NR7s. The outcome is always the very next session.

Three measures of what came next. First, whether the next range was at least twice the NR7’s own range. Second, whether it reached the 20-day average daily range (ADR20, the mean of the last 20 high-minus-low ranges) as of the NR7 day. Third, whether it traded above the NR7 high, below the NR7 low, both or neither.

The samples. SPY daily bars from the Yahoo Finance chart API cover 8,451 days from 1 March 1993 to 24 September 2026 that have at least 20 earlier sessions and a following session. Five large caps, AAPL, MSFT, JPM, XOM and KO, add 33,510 days from 1 February 2000. Every figure below is compared with all days in the same sample, not with zero.

A worked example

Thursday 11 September 2026 was an NR7 on SPY. The high was $766.38 and the low $763.60, a range of $2.78. The six sessions before it ranged $4.70, $6.58, $3.87, $4.56, $3.53 and $3.47. The smallest of those was $3.47, on 10 September, which was itself an NR7, and $2.78 is below it.

The next session did expand, but not through a clean trigger. On Monday 14 September SPY opened at $759.00, already $4.60 below the NR7 low, so a sell order resting just under $763.60 would have filled near the open price rather than at the level. The day ran from $757.93 to $763.52, a range of $5.59, which is 2.0 times the NR7 range and above the $4.24 ADR20 on 11 September. It closed at $760.88.

The wider move came two days later. On 16 September SPY ranged $12.07, the widest session of the month. An NR7 flags that a quiet spell is in place; it does not say which day the spell ends.

22 September repeated the setup. The range was $2.57, under the smallest of the prior six, $3.61 on 17 September. On 23 September SPY fell to $766.50, below the NR7 low of $772.57, never reached the NR7 high of $775.14, and closed at $767.81 on a range of $6.55, about 2.5 times the NR7.

The original data

NR7 days are common. 1,272 of the 8,451 SPY days qualified, 15.1%. If ranges arrived in random order, any day would be the narrowest of seven about one time in seven, 14.3%, so NR7s turned up about as often as chance alone would produce, slightly more in fact. By decade the share ran from 237 of 1,729 (13.7%) in 1993 to 1999 to 279 of 1,691 (16.5%) since 2020.

Measured against itself, the next day expanded often. After an NR7, 350 of 1,272 next sessions (27.5%) covered at least twice the NR7 range. After any SPY day, 836 of 8,451 (9.9%) did. The median next session was 1.49 times the NR7 range.

Measured against a normal day, it usually did not. Only 449 of the 1,272 next sessions (35.3%) reached the 20-day ADR, against 3,557 of 8,451 (42.1%) after any day. The median next range was 0.82 of ADR20 after an NR7 and 0.91 after any day. The NR7 day itself had a median of 0.56 of ADR20, so the next day widened out from a very low base and still usually landed below average.

Breaks came in both directions, often on the same day. The next session took the NR7 high 64.5% of the time and the low 53.2%. Both went on 259 sessions (20.4%), twice the 10.1% after any day, and neither on 33 (2.6%), against 11.4% after any day. The next close finished above the NR7 high 498 times, below the low 374 times and back inside 400 times.

The large caps said the same. Across AAPL, MSFT, JPM, XOM and KO there were 5,010 NR7 days. The next session doubled the NR7 range 21.3% of the time (7.3% after any day), reached ADR20 33.0% of the time (41.8%), and took both ends 19.3% of the time (9.0%). Every SPY NR7 day and its next session is listed in the SPY NR7 file.

Narrow compared with what?

Any narrow day behaves much like an NR7. To test whether the seven-day rule adds something, SPY days with a range under 0.7 of ADR20 that were not NR7 days were counted separately: 1,136 of them. The next session doubled their range 19.3% of the time, reached ADR20 29.8% of the time, and took both ends 17.6% of the time.

So the effect is mostly narrowness itself. A small day is easy to double, and the NR7 filter picks small days. NR7 days that were also inside days (high below the prior high and low above the prior low), 316 SPY cases and a small sample, were a little sharper: the next session doubled the range 30.7% of the time and took both ends 21.8% of the time.

Neither version moved price further in the end. The median next-day close-to-close move was 0.52% after an NR7 and 0.54% after any day.

When it fails

The next day stays quiet. After the 10 September 2026 NR7, SPY’s next session, 11 September, was narrower still at $2.78. Nearly two out of three NR7s were followed by a session that fell short of the 20-day ADR.

Both sides go. On 20.4% of next sessions SPY traded through the NR7 high and the NR7 low. A trader holding the two-sided orders gets filled one way, then stopped on the other side of a range that was tight to begin with, so the stop sits close and is easy to reach.

The open gaps past the trigger. The 14 September 2026 example opened $4.60 through the NR7 low. Orders resting at the level fill at the open, and the tight risk that made the setup attractive disappears on the fill.

Half days make cheap ones. A shortened session produces a small range for reasons that have nothing to do with a buildup of pressure. Three of the seven SPY half days from November 2023 to December 2025 were NR7s, and none of the three was followed by a session that reached the 20-day ADR, a tiny sample but a clear warning.

Clusters dilute the signal. Quiet spells produce NR7s on back-to-back days, as on 10 and 11 September 2026. Each one resets the trigger levels, and a trader following all of them pays for the one that is finally followed by a wide day with the ones that were not.

Related

The inside bar page covers the other contraction day, the one that shares its edges with the day before and combines with NR7 in the tighter setup above. The Bollinger squeeze looks for the same quiet phase using band width over many days rather than one day’s range. And the breakout page explains the move an NR7 is meant to precede, and what separates a break that follows through from one that falls back.

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