Week commencing Monday 31 August 2026

At a glance

Every window that appeared, in the order the indicator produced them.

▲ / ▼ / ∅ are the measured shares of comparable historical windows in which the upper target was reached first, the lower target was reached first, or neither was reached by the session close; n is how many such windows sit behind them. Every reading on this page was taken with the target multiple set to ×0.5, the setting the Atlas IB handbook recommends for NQ and the default in the current build, unchanged since w/c 17 August; the first two weeks on the record ran on Auto. These captures are 5-minute charts, so outcome times are the bar in which the rail was crossed and are good to the bar. The session close is the 16:55 ET bar: a window that reaches neither rail is scored to that bar. No window this week resolved on it.

Two comparisons appear in every capture, and they answer different questions. The box tag grades the window against its own last five sessions (💰 accumulation, ⚖️ normal, 🚀 expansion), while the status table's regime row grades it against the full seventeen-year history, and the two legitimately disagree: Tuesday's hour and Friday's hour were ⚖️ Normal on the box and compressed in the table, Wednesday's half hour was 💰 Accumulation on the box and compressed in the table, Monday's half hour Accumulation and normal, and Thursday's half hour Normal and expanded. With the multiple fixed at ×0.5 the regime no longer moves the rails, so it is reported here but it is not doing anything to the targets; Auto would have set ×1 on the compressed boxes and ×0.25 on the expanded one. The Window strength row read 68% favoured-rail all week, #14 of 91 for the half hour and #11 of 91 for the hour; it describes the window rather than the day.

Two red folder USD releases this week, both marked against their day. ISM manufacturing at 10:00 ET on Tuesday, the minute the half-hour box closed and the middle of the hour's window; both Tuesday windows resolved after it. Non-farm payrolls, average hourly earnings and the unemployment rate at 08:30 ET on Friday, an hour before the window opened. Monday 31 August was a normal trading day.

The 60-minute window · 09:30 to 10:30

The longer window waits for the whole first hour to complete before it commits. It gets a bigger, better-defined range to work from and, at a fixed multiple, a target that sits further away in points.

Mon 31 Aug 2026 · 09:30 +60m · NQ 5-minute

Monday 31 August

Last Wednesday's cell again, and this time the late rally stopped short

∅ neither target reached · the 38% branch

• IB range163.00 pt · avg5 216 · ⚖️ Normal
• Split▲ 13% · ▼ 49% · ∅ 38% (n 435)
• Shorter horizons▼ 2h 31% · 4h 41%
• Target distance▼ 81.5 pt (×0.5)

163 points against a five-day average of 216, ⚖️ NORMAL on both rulers, and the read was 13 / 49 / 38 from 435 sessions: the cell that produced the record's first lean marked against, on Wednesday 26 August, back for its second draw. 49 of the 62 that resolve is 79% of the decided share; magnet on the lower rail, target 81.5 points under the box.

The shape of the day was Wednesday's again in miniature. Price ranged inside the box until mid-afternoon, then rallied from about 15:30 into the close. This time the high, 29,538, was about sixty points short of the upper rail at 29,598.5, and the session low after the box, 29,363.25, was ninety short of the lower one. Neither target by the 16:55 bar.

Two draws of the cell: against, then neither. The 49% it favours has not landed yet. A cell that says 38% reach neither and 13% reach the other side has now done each once, which is unremarkable, and the record notes it and moves on.

Tue 1 Sep 2026 · 09:30 +60m · NQ 5-minute

Tuesday 1 September

The 49 / 11 / 40 cell for a second time, and it paid fifteen minutes after the window closed

🧲✅ favoured side reached first · ~10:45 🔴 ISM Manufacturing PMI · 10:00 ET

• IB range139.00 pt · avg5 189.4 · ⚖️ Normal
• Split▲ 49% · ▼ 11% · ∅ 40% (n 620)
• Shorter horizons▲ 2h 27% · 4h 40%
• Target distance▲ 69.5 pt (×0.5)

ISM manufacturing printed at 10:00 ET, half way through the hour. The box that closed at 10:30 was 139 points, ⚖️ NORMAL against a 189 average and compressed on the table's ruler, and it read 49 up, 11 down, 40 neither, from 620 sessions: the cell Thursday 27 August's hour returned, when it was not reached from a 198-point box. 49 of the 60 that resolve is 82% of the decided share; magnet on the upper rail, target 69.5 points above the box.

Price made the hour's low at about 09:45, climbed through the rest of the window, and kept going. The target went at about 10:45, three bars after the box closed. Magnet and tick.

The same cell comes back on Wednesday from a box 38 points wider and is not reached. Three draws under this lean test now stand at one reached, two not, none against, which is what a 49 / 40 split is allowed to do.

Wed 2 Sep 2026 · 09:30 +60m · NQ 5-minute

Wednesday 2 September

Third draw of the cell in five sessions, seventy points short at four o'clock

∅ neither target reached · the 40% branch

• IB range176.75 pt · avg5 175.85 · ⚖️ Normal
• Split▲ 49% · ▼ 11% · ∅ 40% (n 620)
• Shorter horizons▲ 2h 27% · 4h 40%
• Target distance▲ 88.4 pt (×0.5)

The hour rallied from its first bar to its last, 177 points against a 176 average, as ⚖️ NORMAL as a box gets. 49 / 11 / 40 from 620 for the third time in five sessions. Magnet on the upper rail, target 88 points above the box high.

Price spent the day between the box midpoint and a few points above its high. The best it managed was 29,212.5, at about 16:00, seventy points short of the rail at 29,282.38; the low after the box was 29,106.25. Neither target by the 16:55 bar.

Three draws of the cell now: not reached from 198 points, reached from 139, not reached from 177. The half hour on the same morning read a different, narrower box, abstained, and was marked with a cross within twenty minutes; that is further down.

Thu 3 Sep 2026 · 09:30 +60m · NQ 5-minute

Thursday 3 September

The hour and the half hour drew the same box, and both were paid on the same bar

🧲✅ favoured side reached first · ~11:10

• IB range176.00 pt · avg5 175.8 · ⚖️ Normal
• Split▲ 47% · ▼ 15% · ∅ 38% (n 581)
• Shorter horizons▲ 2h 25% · 4h 36%
• Target distance▲ 88.0 pt (×0.5)

The half hour set both extremes of the morning and the second half hour stayed inside them, so the 60-minute box was the 30-minute box: 176 points, the same high, the same low, the same rails, ⚖️ NORMAL against a 176 average. The hour's table read 47 / 15 / 38 from 581 sessions, 76% of the decided share, and the magnet printed on the upper rail with the target 88 points above the box.

Price dipped back into the lower half of the box at about 10:40, reversed, and ran up through the rail at about 11:10. Magnet and tick, on both windows, on the same bar.

Two identical boxes do not make two independent tests. The hour and the half hour looked up different cells (581 sessions against 455) and leaned the same way, and one move settled both. Count it once if you are counting.

Fri 4 Sep 2026 · 09:30 +60m · NQ 5-minute

Friday 4 September

Payrolls day, an abstain at 24 against 35, and the larger share landed

✅ larger share reached first · ~11:20 · abstained window 🔴 Non-farm payrolls · 08:30 ET

• IB range138.25 pt · avg5 171.35 · ⚖️ Normal
• Split▲ 24% · ▼ 35% · ∅ 41% (n 347)
• Shorter horizons▲ 2h 6% · 4h 15% · ▼ 2h 18% · 4h 29%
• Target distance▼ 69.1 pt (×0.5)

Non-farm payrolls printed at 08:30 ET, an hour before the window opened, and the chart shows what it did to the pre-market. The hour itself was 138 points, ⚖️ NORMAL against a 171 average and compressed on the table's ruler, and once again it was the half hour's box: the second half hour added nothing to either extreme. The read was 24 / 35 / 41 from 347 sessions, the same cell as Monday 17 August's hour, with ∅ the largest share; 35 of the 59 that resolve is 59% of the decided share, short of 62, and both rails stayed grey.

Price fell out of the box after 10:30 and ground lower for the next fifty minutes, crossing the lower rail at about 11:20. The larger of the two grey shares was the side that got there, so the mark is a plain green tick on a window that had offered no lean.

Two draws of this cell, both abstains, both reached on the 35% side: 17 August at about 13:23, and this one. The half hour on the same box leaned long from a different cell and was marked against by the same bar; that comparison is further down.

The 30-minute window · 09:30 to 10:00

The shorter window commits half an hour earlier, from a smaller range, and its target is therefore usually nearer. It is a different measurement of the same morning, not a second opinion on it.

Mon 31 Aug 2026 · 09:30 +30m · NQ 5-minute

Monday 31 August

Last Thursday's abstain cell again, and this time it reached neither

∅ neither target reached · the 17% branch

• IB range130.25 pt · avg5 182.85 · 💰 Accumulation
• Split▲ 33% · ▼ 50% · ∅ 17% (n 347)
• Shorter horizons▲ 2h 18% · 4h 26% · ▼ 2h 38% · 4h 44%
• Target distance▼ 65.1 pt (×0.5)

130 points, 💰 ACCUMULATION against a 183 average and normal on the table's ruler, and 33 / 50 / 17 from 347 sessions, the medium sample bin. It is the cell Thursday 27 August's half hour returned; 50 of the 83 that resolve is 60% of the decided share, two short of the lean test, so both rails were grey again.

Price sat in the lower part of the box and briefly under it, the low at 29,354 about thirty points above the lower rail, and the late rally to 29,538 stopped about forty-five points under the upper one. Neither target by the 16:55 bar, on a cell that gives that outcome 17%.

Two abstains from one cell: a cross on 27 August, neither here. Nothing was offered on either morning, and nothing is claimed for either.

Tue 1 Sep 2026 · 09:30 +30m · NQ 5-minute

Tuesday 1 September

Forty-seven against thirty-nine, short of a lean, and the larger share landed on the ISM move

✅ larger share reached first · ~10:35 · abstained window 🔴 ISM Manufacturing PMI · 10:00 ET

• IB range117.75 pt · avg5 149.7 · ⚖️ Normal
• Split▲ 47% · ▼ 39% · ∅ 14% (n 367)
• Shorter horizons▲ 2h 32% · 4h 41% · ▼ 2h 29% · 4h 37%
• Target distance▲ 58.9 pt (×0.5)

The half-hour box closed at 10:00, the minute ISM manufacturing was released. 118 points, ⚖️ NORMAL, and a table that expected a resolution, ∅ 14%, but could not say which way: 47 up, 39 down, 55% of the decided share. Both rails grey.

Price climbed from the close of the window without a meaningful pullback and crossed the upper rail at about 10:35. The larger share was the upside, so a plain green tick printed on an abstained window. The hour, which had the release inside its box, leaned long and was paid ten minutes later by the same move.

A 47 / 39 split is not a lean and the indicator did not call it one. The grade records that the larger of two shares got there first, and that is all it records.

Wed 2 Sep 2026 · 09:30 +30m · NQ 5-minute

Wednesday 2 September

The 38 / 48 / 15 cell for a third time, a third abstain, and a cross inside twenty minutes

❌ larger share not reached first · ~10:20 · abstained window

• IB range101.75 pt · avg5 145.9 · 💰 Accumulation
• Split▲ 38% · ▼ 48% · ∅ 15% (n 371)
• Shorter horizons▲ 2h 27% · 4h 32% · ▼ 2h 38% · 4h 44%
• Target distance▼ 50.9 pt (×0.5)

The smallest box of the fortnight: 102 points, 💰 ACCUMULATION against a 146 average and compressed on the table. The read was 38 / 48 / 15 from 371 sessions, the cell of Tuesday 11 August and Friday 28 August, and for the third time 48 of the 86 that resolve was 56% of the decided share and the window abstained. Target 51 points, the nearest of the week.

Price rallied out of the box on the first bars after 10:00 and crossed the upper rail at about 10:20. The larger share was the downside, so the mark is a ❌. The hour on the same morning, whose box was 75 points taller and whose rails sat further out, leaned long and reached neither.

Three draws of the cell, three abstains: neither, cross, cross. Nothing was offered on any of the three mornings. The ❌ on an abstained window is the record's most frequent cross, and it is a grade of the larger share, not of a call.

Thu 3 Sep 2026 · 09:30 +30m · NQ 5-minute

Thursday 3 September

The same box as the hour, a different cell, the same lean, and the same bar

🧲✅ favoured side reached first · ~11:10

• IB range176.00 pt · avg5 134.45 · ⚖️ Normal
• Split▲ 56% · ▼ 28% · ∅ 16% (n 455)
• Shorter horizons▲ 2h 37% · 4h 48%
• Target distance▲ 88.0 pt (×0.5)

176 points in the first half hour, ⚖️ NORMAL on the box against a 134 average and expanded on the table. The second half hour did not extend it, so the hour that closed at 10:30 had the same box and the same rails. The half hour's cell was 56 / 28 / 16 from 455 sessions, 67% of the decided share; magnet on the upper rail, target 88 points above the box.

Price dipped back into the lower half of the box at about 10:40 and then ran up through the rail at about 11:10. Magnet and tick, on the same bar as the hour's.

Two cells, one box, one bar. The half hour's table was more decided than the hour's (56 against 47 on the upside) and its ∅ share was less than half the hour's (16 against 38), because the two windows are compared with different histories even when they measure the same range. On this morning the difference did not matter.

Fri 4 Sep 2026 · 09:30 +30m · NQ 5-minute

Friday 4 September

The strongest cell on the page, leaning long on payrolls day, and the second lean marked against

🧲❌ opposite side reached first · ~11:20 🔴 Non-farm payrolls · 08:30 ET

• IB range138.25 pt · avg5 133.45 · ⚖️ Normal
• Split▲ 65% · ▼ 16% · ∅ 19% (n 635)
• Shorter horizons▲ 2h 48% · 4h 57%
• Target distance▲ 69.1 pt (×0.5)

Payrolls had printed an hour before the open. The half hour was 138 points, ⚖️ NORMAL against a 133 average, and the hour that followed added nothing to it, so both windows shared the box and the rails. The half hour's cell was 65 / 16 / 19 from 635 sessions, 81% of the decided share, the strongest lean of the fortnight, and the cell Wednesday 26 August's half hour was paid from on the closing bar. Magnet on the upper rail, target 69 points above the box.

Price fell out of the box after 10:30 and kept falling. The lower rail went at about 11:20, and the mark is a magnet-and-cross: the second Atlas-mode lean on the record to resolve against, seven sessions after the first.

The hour, reading the same box through a different cell, had abstained at 24 against 35 and was marked with a plain tick by the same bar. Two cells, one box, one bar, and the more confident of the two was the one that was wrong. Two draws of 65 / 16 / 19: reached, then against.

Two mornings on which the hour and the half hour were the same box

On Thursday and Friday the second half hour stayed inside the first half hour's range. The 60-minute box and the 30-minute box had the same high, the same low and, at a fixed ×0.5, the same rails. Four windows, two boxes, two bars.

Friday is the second Atlas-mode lean on the record to resolve against, seven sessions after the first. Across five weeks and twenty-nine Atlas-mode windows the count is now fourteen paid, thirteen unreached, two against. Both came from cells that put 13% and 16% on the side that landed; one was its cell's first draw, the other its second, after a first that paid. Neither is surprising, and the sentence the earlier pages carried, that no lean had yet been marked against, is retired.

The repeat cells this week were mostly the hour's. 49 / 11 / 40 from 620 sessions came back on Tuesday and Wednesday after Thursday 27 August, and under this lean test stands at one reached, two not, none against; it had also appeared on 4 August under the earlier build's lean test, where it abstained, and that draw is not counted with these. 13 / 49 / 38 from 435 returned on Monday after last Wednesday's cross and reached neither. On the half hour, 33 / 50 / 17, 38 / 48 / 15 and 65 / 16 / 19 all made their second or third appearance, and the abstained cells were graded on their larger share as they always are.

What this week actually shows

Five things, from a week that leaned long and was settled early.

1. Six leans, three paid, one beaten. Five long leans and one short. Tuesday's hour, Thursday's hour and Thursday's half hour reached their targets, all before 11:15; Monday's and Wednesday's hours reached neither; Friday's half hour was marked against. The running count across five weeks is 29 Atlas-mode windows: 14 paid, 13 unreached, 2 against. Twenty-nine is still far too few to call a property. It is a count, and it is published as one.
2. Two boxes served four windows. On Thursday and Friday the hour did not extend the half hour, so the two durations shared rails and were settled by the same bar. Thursday's pair agreed and both paid; Friday's pair disagreed, the abstaining hour was graded right on its larger share and the leaning half hour was graded wrong. Count each morning once.
3. Every resolution came early. Seven of ten windows resolved, and all seven did so between 10:20 and 11:20. The three that reached neither, Monday's two windows and Wednesday's hour, all got within seventy points of a rail and no nearer. Nothing this week was decided after lunch.
4. Four abstains, graded four ways. Monday's half hour reached neither, Tuesday's half hour and Friday's hour ticked on their larger share, Wednesday's half hour crossed. None of the four offered a lean, and the marks record what the larger share did. Six of the eight crosses in the fifty-window record now sit on abstained windows; the other two are the two leans marked against.
5. Fifty windows settle nothing. Three of six leans paid, on cells that said 47, 49, 49, 49, 56 and 65. A week cannot confirm or refute a frequency measured over hundreds of sessions, and a second miss against a lean cannot either. What the record can do is accumulate in public, the bad weeks beside the good ones, which is why this one goes up.