A plain-English guide to the no-call band: what a plan learns when the print lands inside it, and what it does when the day belonged to something else.

Written Thursday November 5, 2026, accompanying the Wednesday November 4 analysis note, which graded the October ISM Services PMI against a band of 54.0 to 55.8 and got a print of 54.5. The plan had set that band deliberately narrow, a point either side of September's 54.9, because the wider bands of earlier plans had left 17 of 33 scorecard lines as no calls. It produced a no call anyway, and on a day when Brent fell 8.7 percent and gold rose 3.4 percent on causes the plan never mentioned. This guide explains what a no-call band is, how its width gets chosen, what a no call does and does not tell you, and what a pre-registered plan does when the session belongs to something it did not register.

What the band is

Every plan on this site pre-registers a reaction map: a scheduled release, two fixed levels around its prior, and a conditional statement for each side. Above the upper level is "firm" and the map says yields close higher than the previous day; below the lower level is "soft" and yields close lower; between the two levels the map says nothing. That middle zone is the no-call band. The levels are written as fixed numbers rather than as a distance from the prior, so that a revision to the prior cannot move them (data revisions covers why). The grade, at the close, is against the previous session's daily close on the 5- and 10-year yields (Yahoo ^FVX and ^TNX), and nothing is regraded afterwards.

The band exists because a conditional statement needs an antecedent that is actually true or false. "If the print is a surprise, yields close on the side of the surprise" is a hypothesis; "if the print is 54.5 against 54.9, yields close lower" is not, because a 0.4-point change in a diffusion index is not a surprise by any reading of that series. The band is the plan's definition of "surprise", written down before the print.

How the width gets chosen, and the trade-off

The width is set from the series' own recent noise, and the plan states the rationale in one sentence so a reader can check it. The ISM services plan put the lines at 53.9 and 55.9 because the last two monthly changes it could source were -0.5 and +1.3 (the latter from a secondary calendar), so a one-point move was "about the size" of an ordinary month. The Thursday November 5 plan put initial claims lines at 179,000 and 209,000, 15,000 either side of the prior 194,000, because the eight weekly prints since early September spanned only 13,000 (194,000 to 207,000, FRED ICSA); a print outside the band is outside the range the series has occupied for two months.

The trade-off is unavoidable. A band that is too wide never fires, and a plan that cannot be wrong is not a plan; before Wednesday, 17 of the 33 graded lines on the scorecard were no calls, about half. A band that is too narrow fires on noise, and then the plan grades whether yields moved with a 0.4-point wobble that no one traded; the hits and misses that come out of that are coin flips dressed as evidence. Wednesday's lesson is that one point was still too wide for the ISM services headline, or, put the other way, that the series does not move a point often enough to be a good graded leg: its last three monthly changes were -0.4, -0.5 and +1.3, so a one-point band holds it two months in three. The honest response is to say so in the grade, keep the band for the series where the noise argument works, and stop treating the ISM headline as a leg that will usually fire.

What a no call tells you

• Nothing about the hypothesis. The conditional's antecedent did not fire, so the consequent was never tested. Wednesday's yields fell 5.4 and 5.8 basis points; had the plan called soft, those would have been hits, and had it called firm, misses. Neither happened, and it would be wrong to count the day as evidence for or against "surprises move yields".
• Something about the band. Each no call is one more observation that the series stayed inside the width chosen. Enough of them, and the width was wrong for that series, which is a finding about the plan, not the market.
• Not a miss, and not a dodge either. The scorecard records no calls as their own result, separately from hits and misses, so that the tally cannot be flattered by them; after Wednesday the line count is 7 hits, 9 misses, 19 no calls, 2 void and 1 outcome. A plan that wanted to look good would widen every band. The count is published so the reader can see whether it has.

When the day belonged to something else

Wednesday's session was decided by two things the plan did not register: an 8.7 percent fall in Brent, which NBC News and Bloomberg attributed to traders repricing the chance of the Iran war ending after the midterm elections, and a 3.4 percent rise in gold, which Reuters attributed to a falling dollar after the same elections (sources and quotations in the analysis note). The ISM release itself moved the 5-year up 2.1 basis points for about half an hour and was then swamped. What does a pre-registered plan do with that?

1. It does not regrade. The grade is against the fixed levels on the daily close, whatever else happened. A plan that excluded "unusual" days after seeing them would be choosing its own sample. The Nov 4 plan wrote this down before the fact: anything election-related "is carried in the grade, accepted rather than adjusted for". The Nov 5 plan says the same of the Bank of England decision that lands at 7:00 AM ET, inside its window.
2. It leads the close note with what actually moved. The grade is the grade, but the note is a record of the session, and a record that opened with a 0.4-point survey change on a day oil fell nine percent would be misleading by arrangement. The analysis note leads with Brent and gold, with named outlets and dates for the attributed cause, and keeps them out of the grade.
3. It records the event window. Thirty minutes of yield data around the release, kept separate from the close, show how much of the day the release can plausibly own. On Wednesday the window went up and the day went down; on Tuesday November 3 a third of the 5-year's rise was in place before the JOLTS release. The window never changes a grade; it changes how much the writer is allowed to claim for it. Event windows is the long version.
4. It names the cause only when a named outlet has. If no Reuters, Bloomberg, AP or equivalent story can be found that dates the move and gives a reason, the note says "cause not checked" and leaves it there. A plausible story written by the analyst is the failure mode this site exists to avoid.
5. It uses the conflict rule only between registered legs. When a plan has two legs pointing opposite ways, its conflict rule says which wins or declares the line void. An unregistered driver is not a leg; it cannot void anything, and a hit or a miss that happens to coincide with it still counts. The Oct 7 note voided a 10-year call because the FOMC minutes and a 10-year auction pulled against each other; it did not void the dollar and gold calls, which were graded as hits even though, as that note says, nearly all of their moves came before the minutes were released.

A checklist for reading any graded note

• Find the band and the one-sentence width rationale in the plan. If the rationale is missing, the band is arbitrary.
• Check whether the print landed inside it. If it did, nothing below is evidence about the hypothesis.
• Read the event window before the close. A hit whose move was in place before the release is a hit the plan does not get to explain.
• Look for a named outlet and a date behind any stated cause. "Traders said" without a source is a guess.
• Count the no calls on the scorecard against the total. If they grow faster than the lines, the bands are drifting wider.

Where this shows up in practice

• November 4 (Wednesday): ISM services 54.5, two no calls, Brent down 8.7 percent on a sourced driver (the live example)
• October 30 (Friday): ECI 0.8 percent inside the band, the 5-year fell 8 basis points before the release
• November 3 (Tuesday): JOLTS past the firm line, two hits, a third of the move before the release

Related reading

• The paired analysis: Wednesday November 4
• Event windows
• Data revisions, and why thresholds are fixed levels
• ISM Manufacturing vs Services
• Surprise indices

Nothing on this site is investment advice or a recommendation to trade. Most retail traders lose money.