Nine of the ten windows leaned down, Atlas mode fired seven and was paid on six, and one 10:40 bar scored Monday both ways

The week at a glance

Every number here is transcribed from the capture of that window. Times are read off the 5-minute bars, so an outcome time is the bar in which the rail was crossed.

Regime is the table's tag, which measures this box against the whole history; the tag printed on the box itself measures it against the last five sessions, so the two legitimately differ. They are the same three buckets named twice, compressed against accumulation, expanded against expansion, normal against normal. This week eight of the ten boxes land in the same bucket on both rulers; Monday's half hour and Thursday's hour are the two that differ, each compressed against the whole history and normal against the last five sessions. Mult is an output of the table's regime, never a setting, and every window here reads auto. The four compressed boxes drew a full multiple, the two normal ones a half and the four expanded ones a quarter, which is why Friday's 48.3 point hour carries a 12.1 point target. Targets print rounded to a tenth and the rails are drawn from the unrounded figure, so a printed target does not always reproduce a printed rail exactly: Monday's hour is 12.45, Wednesday's half hour 5.975 and Friday's windows 10.65 and 12.075. Nothing was touched between any two windows. The three-way shares do not all sum to 100, because each is rounded to a whole percent: Monday's hour and both of Thursday's windows print 101. Charts are 5-minute, so an outcome time is the bar in which the rail was crossed and is good to the bar; the session close is the 16:55 ET bar on all five days. The tick and cross in the outcome column mark whether the favoured rail was the one that went, for calls and abstains alike; only the seven Atlas rows count towards the Atlas score. Monday's tick and cross are one bar, 10:40: the same drop paid the half hour's short and crossed the hour's long.

Morning by morning

The 30 minute window on the left, the 60 minute on the right. Click any chart to enlarge.

Monday 28 September

One 10:40 bar paid the half hour's short and crossed the hour's long

30m Atlas mode ✓ · 60m Atlas mode ✗

The two windows read the morning in opposite directions and both cleared both gates. The half hour read 23 / 49 / 28 from 215 sessions, 68 percent of the decided share and 72 percent resolved, and fired short. The hour read 49 / 19 / 33 from 308 sessions, 72 percent of the decided share and 67 percent resolved, and fired long, the only window of the ten that leaned up. The hour's three shares print as 101 because each is rounded to a whole percent.

The half hour's box ran 4173.5 to 4187.0. Price dipped below it to 4168.3 on the 09:35 bar, then climbed through its top to 4193.2 on the 09:55 bar, 7.3 points short of the half hour's upper rail at 4200.50, and that bar set the hour's box high. The hour's box, 4168.3 to 4193.2, is 24.9 points and normal on both rulers, so Auto drew a half multiple and a 12.45 point target, printed as 12.4, which put its lower rail at 4155.85, 4.15 points below the half hour's at 4160.00.

From the 10:00 bar price worked lower, and the 10:40 bar ran from 4164.3 to 4155.0. That took the half hour's lower rail by 5.0 points, an hour and ten minutes after its box closed, and the hour's by 0.85 of a point, forty minutes after its box closed. One bar, one Atlas hit and one Atlas cross. Price went on to 4143.0 on the 10:55 bar. The half hour's cell drew once before, on 8 September, fired short and was paid then as well; the hour's cell is new to this record.

• 30m box13.5 pt · compressed
• 60m box24.9 pt · normal
• 30m three-way23 / 49 / 28
• 60m three-way49 / 19 / 33
• 30m modeAtlas, short
• 60m modeAtlas, long
• Outcome30m down rail 10:40 · 60m down rail 10:40
• Durationsdisagreed

Tuesday 29 September

Both windows abstained with the down side larger, and both went to the upper rail, the half hour at 10:15 and the hour at 16:30

Both windows abstained and the other rail landed

Both windows leaned down and neither would call it. The half hour read 35 / 43 / 22 from 270 sessions: 78 percent of comparable sessions resolved, but the down side was only 55 percent of the decided share against a 62 percent gate, so it abstained. The hour read 18 / 24 / 58 from 247 sessions, 56 percent of the decided share and 42 percent resolved, under both gates, and abstained as well. Both then went to the upper rail, the smaller share. No call was made, so no call was crossed, but the favoured rail was the wrong one twice.

The half hour's box ran 4182.5 to 4192.0, 9.5 points. Price edged above it from the 09:30 bar, the 10:05 bar stopped 0.1 of a point under the upper rail at 4201.50, and the 10:15 bar took it on a high of 4204.4, forty-five minutes after the box closed. The cell, 35 / 43 / 22 out of 270, has now drawn three times on this record and abstained each time: the upper rail went on 15 September, the lower on 17 September and the upper again today.

The hour's box topped out at 4196.8 on the 09:50 bar, and its upper rail sat at 4211.10. Price reached 4207.3 on the 10:20 bar, then sank for three and a half hours to 4174.4 on the 13:50 bar, 6.2 points above the lower rail at 4168.20. The 14:00 bar ran 20 points in five minutes and the afternoon kept climbing: the 15:55 bar stopped 0.4 of a point short of the upper rail and the 16:30 bar went through it on a high of 4211.8, six and a half hours after the box closed, the slowest resolution of the week. Both of Tuesday's cells abstained; only the hour's is new to this record.

• 30m box9.5 pt · compressed
• 60m box14.3 pt · compressed
• 30m three-way35 / 43 / 22
• 60m three-way18 / 24 / 58
• 30m modeabstain, direction gate at 55
• 60m modeabstain, both gates
• Outcome30m up rail 10:15 · 60m up rail 16:30
• Durationsagreed

Wednesday 30 September

Two expanded boxes, two quarter multiples, and the half hour's short paid on the second bar after its box

Both durations Atlas mode ✓🔴 08:30 Core PCE m/m, Final GDP q/q

Both windows fired short off wide boxes. The half hour read 17 / 81 / 2 from 215 sessions, 83 percent of the decided share and 98 percent resolved; the hour read 14 / 67 / 19 from 203, 83 and 82. Both boxes topped out on the 09:00 bar at 4246.5 and both are expanded on both rulers, so Auto drew a quarter multiple each time: the 23.9 point half hour carried a 6.0 point target and the 39.2 point hour a 9.8 point one.

The half hour's lower rail at 4216.63 went on the 09:35 bar, the second after its box closed, on a low of 4214.6, the fastest resolution of the week. Its cell, 17 / 81 / 2 out of 215, is last Friday's half hour drawn again three sessions later, and its third draw on this record: short every time, crossed on 25 August, paid on 25 September and paid today. The 81 percent down share equals the largest stated favourite the gold tables have printed here, a figure now drawn six times.

The hour's box kept falling to 4207.3 on the 09:50 bar. After it closed, price held between 4200.4 and 4215.0 for half an hour, and the 10:30 bar dropped to 4196.8, 0.7 of a point through the lower rail at 4197.50. Price went on to 4178.0 on the 13:45 bar. The hour's cell is new to this record.

• 30m box23.9 pt · expanded
• 60m box39.2 pt · expanded
• 30m three-way17 / 81 / 2
• 60m three-way14 / 67 / 19
• 30m modeAtlas, short
• 60m modeAtlas, short
• Outcome30m down rail 09:35 · 60m down rail 10:30
• Durationsagreed

Thursday 1 October

The half hour's 09:40 hit set the hour's floor, and the hour then traded both sides of its box without reaching either rail

30m Atlas mode ✓ · 60m abstained, no target hit

The half hour read 22 / 70 / 9 from 215 sessions, 77 percent of the decided share and 92 percent resolved, and fired short. Its 14.8 point box was normal on both rulers, so the multiple was a half and the target 7.4 points, with the lower rail at 4189.70. The 09:30 and 09:35 bars dipped below the box without reaching the rail, and the 09:40 bar, ten minutes after the box closed, took it on a low of 4188.8. The cell drew once before, on 13 August, fired short and was paid then too. Its three shares print as 101 because each is rounded to a whole percent.

The same 09:40 bar set the hour's box low, so the hour runs 4188.8 to 4211.9, 23.1 points, compressed on the table's ruler and drawn at a full multiple. It read 12 / 28 / 61 from 265 sessions: 71 percent of the decided share, clear of the direction gate, but only 40 percent resolved against a 50 percent gate, so it abstained with the down side as the larger share. These shares print as 101 as well.

With a 23.1 point target either side, the hour's rails sat at 4235.00 and 4165.70. Price went below the box to 4179.4 on the 11:10 bar, 13.7 points short of the lower rail, and above it to 4217.6 on the 13:30 bar, 17.4 points short of the upper, and the 16:55 bar traded between the rails. It was the only window of the ten to reach nothing. Its cell, 12 / 28 / 61 out of 265, abstained on 11 August and 8 September as well, and the lower rail went both of those times.

• 30m box14.8 pt · normal
• 60m box23.1 pt · compressed
• 30m three-way22 / 70 / 9
• 60m three-way12 / 28 / 61
• 30m modeAtlas, short
• 60m modeabstain, decided gate at 40
• Outcome30m down rail 09:40 · 60m no target
• Durationsagreed

Friday 2 October

The week's widest boxes, two quarter multiples, and two Atlas shorts paid twenty minutes apart

Both durations Atlas mode ✓🔴 08:30 Average Hourly Earnings, NFP, Unemployment Rate

Both boxes topped out on the 09:00 bar at 4255.9. The half hour's ran down to 4213.3 on the 09:20 bar, 42.6 points; the hour's to 4207.6 on the 09:55 bar, 48.3 points, the widest box of the week, though not of the record. Both were expanded on both rulers, so both drew a quarter multiple, for 10.6 and 12.1 point targets. The half hour read 30 / 67 / 3 from 268 sessions, 69 percent of the decided share and 98 percent resolved; the hour read 20 / 62 / 18 from 262, 76 and 82. Both fired short, and both cells are new to this record.

The half hour's lower rail at 4202.65 took an hour. The 10:25 bar's low of 4202.9 stopped a quarter of a point above it, and the 10:30 bar went through on a low of 4198.2. That low was 2.7 points above the hour's rail at 4195.53, and the hour waited four more bars: the 10:50 bar ran from a 4200.5 high to 4186.8, fifty minutes after its box closed. Price went on to 4153.8 on the 11:55 bar.

Four of the week's seven Atlas calls drew a quarter multiple, Wednesday's and Friday's, and all four were paid. A quarter multiple puts each rail a quarter of the box beyond it, and a target that close is reached more often by construction. It is not a stronger result.

• 30m box42.6 pt · expanded
• 60m box48.3 pt · expanded
• 30m three-way30 / 67 / 3
• 60m three-way20 / 62 / 18
• 30m modeAtlas, short
• 60m modeAtlas, short
• Outcome30m down rail 10:30 · 60m down rail 10:50
• Durationsagreed

What the week actually showed

• Nine of the ten windows leaned down, the most one-sided week on the gold record. The previous high was last week's eight. Every half hour leaned down, and Monday's hour was the only window that leaned up. The mean stated favourite across the ten was 54, against the 49 the record averages, which says the tables leaned harder than usual, not that they were right more often. None of that is a property of the indicator; it is a description of five sessions in which gold mostly fell in the morning.
• Atlas mode fired seven times and was paid on six, the most Atlas hits in a week this record has drawn. The previous high was four, on the weeks of 10 August and 21 September. Seven calls is not a record, the week of 31 August fired eight, but all seven resolved. Six were short and all six were paid, four of them on quarter-multiple targets that sit close to the box by construction; the one long, Monday's hour, was crossed on the 10:40 bar that paid Monday's half hour. The three abstains split by cause: Tuesday's half hour on the direction gate at 55 percent of the decided share, Thursday's hour on the decided gate at 40 percent resolved, and Tuesday's hour on both, at 56 and 42.
• Nine of ten reached a rail, and the one that reached nothing was Thursday's hour. Nine equals the count of three earlier weeks, 17 August, 24 August and 7 September. The fastest was Wednesday's half hour on the 09:35 bar, the second after its box; the slowest was Tuesday's hour on the 16:30 bar, six and a half hours after its box closed. The favoured rail went first on six of the nine. Of the three that went the other way, Monday's hour was a call, and both Tuesday windows were abstains that leaned down and went up. Thursday's hour stopped 13.7 points short of its lower rail and 17.4 short of its upper, and is scored as no target hit.
• Five of the ten cells had drawn before. Monday's half hour is 8 September's, 23 / 49 / 28 out of 215, fired short and paid both times. Thursday's half hour is 13 August's, 22 / 70 / 9 out of 215, the same story. Wednesday's half hour is last Friday's, 17 / 81 / 2 out of 215, now drawn three times, short every time, crossed once and paid twice. Tuesday's half hour, 35 / 43 / 22 out of 270, has abstained on all three of its draws and gone upper, lower, upper. Thursday's hour, 12 / 28 / 61 out of 265, has abstained on all three of its draws; the lower rail went on the first two and nothing went on the third. The other five cells are new to this record. A cell drawn more than once with different outcomes is what a stated share looks like when it is a frequency and not a forecast.
• Four windows drew a quarter multiple, the second most in a week on this record after the six of 31 August. Wednesday's and Friday's four boxes were all expanded on both rulers. Wednesday's half hour carried a 6.0 point target, the same figure the table printed on 26 August, and only last Friday's 5.6 is smaller on this record. All four fired short and all four were paid. A smaller target is reached more often by construction; it is not a stronger result.
• Eighty windows over eight weeks: sixty-three reached a rail, the favoured rail won forty of those, Atlas mode fired forty-three times and was right on twenty-six of the thirty-six that resolved. These are published because the record has to accumulate somewhere, and eighty races remain far too few to mean anything. The place to judge calibration is still the indicator's own realised versus expected rows, which this week ran 65% to 70% against a 66% to 71% expectation on the general hit rate, every z between -0.7 and +0.7, and 63% to 77% against a 73% to 75% expectation on the Atlas rate, every z between -1.1 and +0.7.
• Gold and the index book read the same five mornings differently. On the NQ tables Atlas mode fired on six of the ten windows; the gold tables fired on seven. Two instruments, two window times, five sessions and ten windows each, and the only thing that follows from it is that the gates were reading different tables. It is recorded because both pages are published from the same week.