A Wide Spread 7 (WS7) day occurs when a session’s high-to-low range is the widest of the past seven sessions. In Toby Crabel's framework, narrow days signal contraction and wide days signal expansion. Traders typically use the narrow day as a setup for a trade at the next open. This study reverses the question: what happens after a WS7, and is the wide day itself tradable?
WS7 (gold-colored bars) in Natural Gas (NG). Oct 21, 2026: Widest range in 19 trading days.
It addresses four questions: how wide a WS7 is, in both range and closing position; whether it breaks out of a narrow day; what happens in the following session across markets; and how long runs of wide days persist. The main finding is that the next session has a smaller range roughly five times in six, while its close does not follow the wide day’s direction. WS7s are not rare: they occur in 14.4% of sessions, or about one in seven, roughly the rate expected if ranges were randomly shuffled. The WS7 itself is not an unusual signal. The recurring contraction that follows is the more significant finding.
WS7 (gold-colored bars) in the Mexican Peso (USD/MXN). Oct 21, 2026: Widest range in 8 trading days.
The Day Itself
A WS7 is a large day and an ordinary kind of large. The median is 1.71 times the day before it, 1.59 times the average true range of the previous fourteen sessions, and 1.53% of the price. A WS4 that never becomes a WS7 is milder, 1.41 times the prior day and barely over its recent average range. The WS7 is the widest session in a median of fourteen days. Two in three are the widest in twenty sessions or fewer. One in eight is the widest in more than sixty, and that tail is the climax, not the typical bar.
The WS4 row includes the WS7 days.
Up and down split evenly. The up day closes a median of +0.91%. The down day closes a median of −0.96%. The next session's average close is −0.04% after the up day and +0.10% after the down day. The range gives way. The price barely moves. The median giveback of the wide day's close-to-close thrust is 8% of that thrust, because the next day is too small to give the move back and too aimless to extend it.
Where It Comes From
The cycle is the useful part, and it runs in one direction. A tight day is followed by a larger day. A wide day is followed by a smaller one. That is Crabel's Contraction/Expansion Principle, the same cycle Larry Williams calls the Natural Cycle of Range Change.
Small Range, Then What: Same 21 markets and the same sessions.
Green: the next range is larger at least 80% of the time. Rose: the expansion has already printed.
An NR7 is followed by a larger range 82.8% of the time, at 1.46 times the narrow day. An IDnr4 is next, 81.4% and 1.44 times. An NR4 is 76.3% and 1.34 times. An ordinary day is a coin flip, 0.99 times yesterday. Futures, the majors, and the volatile pairs give the same ratios. The strongest cell is an IDnr4 on the futures: larger the next day 85% of the time, 1.48 times, on 254 sessions. Connors and Raschke built the inside-day (ID) and NR4 trade on that same fact: the coil is the setup, and the break is the next session.
That larger day is not, as a rule, a WS7. The session after an NR4 is a WS4 on 25.1% of days. The base rate is 24.8%. The session after an NR7 is a WS7 on 12.7% of days, against 14.4% on any day. The expansion is real, and it is modest. It is bigger than the coil. It is not the widest week of the month, and the coil does not make that outcome more likely. The definitions used here are his: a WS4 is the unique widest of four sessions, a WS7 the unique widest of seven, the mirror of NR4 and NR7.
So the setup is the day before. The WS7 is the result. Only about half of WS4 days were even preceded by an NR, an NR4, or an inside day (ID), which is how often those bars occur anyway. A wide day is not the signature of an IDnr4 breaking. It is just a wide day.
The Next Session
The session after a WS7 is narrower 83.9% of the time, and that range is 0.65 of the spike. Crude Oil does it on 90.9% of its 66 cases. The Yen (USD/JPY), the Swedish Krona (USD/SEK), and the Norwegian Krone (USD/NOK) do it on about 88%. Nasdaq keeps its range the most and is still narrower 77% of the time. Another WS4 the next day is slightly less common than on an ordinary day, 21.5% against 24.8%. A second outside day (OD) falls to 3% from an 11% base. The expansion does not immediately engulf again.
Rose: the close continues 40% or less.
The close is the part that tempts a trade, and it does not pay. The next close follows the WS7 only 46.2% of the time. Two WS7 days in three finish in the outer quarter of their range. Those follow through 45.1% of the time. A close on the extreme is not a reason to hold.
The green cells on the right of that table are the exceptions, and they are small. Crude's up-close continued on 70% of 30 days. Bitcoin's down-close continued on 62% of 37, and the Krona's on 65% of 34. The Yen failed in both directions, 38% and 30%. Natural Gas printed the largest average up-close in the set, +6.86%, and the next session still went with it less than half the time. Same range cycle. Different next close. None of those cells is large enough to become a system.
An inside day (ID) does show up more often the next session, 28% after a WS7 against a 14% base. Part of that is arithmetic. A wide bar is an easy container to trade inside. It is still the contraction shape that becomes more common. An NR4 the next morning does not. It stays near its ordinary one-in-four rate, and when it does appear the spike is still inside its four-day window, so "narrowest of four" is easy. That is not yet a fresh coil.
When a WS7 Streaks
A streak here means consecutive sessions that are each a WS7. The chance of another one does not grow as the run gets longer, and it does not shrink until the sample runs out. A second WS7 arrives 16.2% of the time. A third, 2.6%. A fourth, 0.46%. A random day is a WS7 14.4% of the time, so the label has almost no memory. About five of these days in six are the end of the run. Once one has printed, the expected extra length is 0.2 sessions.
Four in a row happened six times in two years, across all 21 markets. Five never did. The session after the fourth day was narrower than that fourth day in every case. In five of the six, the next five sessions all stayed under it. Nasdaq, in the first week of November 2025, is the one that made a new high range later that week. None of the six was a four-day trend. The close flips inside the run. Gold and the New Zealand Dollar against the Yen (NZD/JPY) rose for two sessions and fell for two, while the range kept making highs. The expansion is in the range. The fourth day is where it stops.
What to Do with All This
An NR7 or an IDnr4 says the next range expands against the coil about five times in six, by roughly half again. That is the session to work. If today has already printed the WS7, the expansion you were waiting for has been spent. Do not arm another breakout because the range was wide. Expect something near two-thirds of that range on the next session. Stand aside, or trade a much smaller day.
Do not chase the close. Direction is a coin flip, slightly against the wide day, on futures and on FX together. Do not fade the whole move either. The next session only walks back a small fraction of it. Crude's up-day is the one continuation worth a look, and it is thirty days. The Yen is the close that fails most often. Neither changes the rule for the other nineteen markets. If a second WS7 prints, you are in the 16%. Take something off if you are still long or short from the coil. A third is one start in forty. A fourth is a list of six, and the next session contracted in all six. There was no fifth in this study window to wait for.
The next setup is the next real contraction, once the spike has left the lookback. An inside day (ID) the next morning is a pause inside a wide bar, not automatically a new NR4. A fresh NR4 or IDnr4 after that spike has dropped out of the window is the coil again. Then the cycle applies, and the trade is that next session, not the wide one you just had.
See also:
Crabel Pattern Occurrence Stats for 7 Futures and 14 FX pairs, October 2024 to date.