NY AM, week commencing Monday 24 August 2026

At a glance

Three scores, because the indicator makes three separable claims.

The four share columns are the indicator's own figures, read off the status table at the session open, and they sum to 100 on every row. Gold marks the branch the tables rated highest. London printed below its own average range on Monday, Wednesday and Friday, and Friday's 76 points is the tightest prior session in the record; Tuesday and Thursday were above. Two red folder USD releases fell inside the session this week: Core PCE and the GDP revision at 08:30 ET on Wednesday, an hour in, on the bar that took the low, and the Warsh speech from Jackson Hole with the benchmark payrolls revision at 10:00 ET on Friday, half an hour after both levels had already gone. Both are marked against the day. Times are the 5-minute bar in which the level was crossed.

The five NY AM sessions

Left, the session opening at 07:30. Right, the same chart at 11:30 with the outcome latched.

Monday 24 August

The high in the second bar, then a 170-point break of the low at the cash open

First sweep ✓ · both sides, the 41% favourite ✓

London printed 133 points, below average, swept its own low and left price above EQ, and the table came back 34 / 25 / 41 / 0 from 528 sessions, the fourth time the record has drawn it. The high, 25 points above the open, went in the second bar of the session at 07:35.

Price then held the top half of London's range for two hours and did nothing, until the cash open. The 09:30 bar fell straight through the London low, latched taken at 09:30, and the fall ran to 28948, 170 points beneath the level, before a recovery that got back to the low itself by the close.

Three from three: first sweep on the high at 58%, both sides the 41% favourite, and both marginals taken, the 75% high and the 66% low. This bin has now resolved both sides on three of its four draws; the exception was Tuesday 11 August, which took the low and stopped.

• London range133 pt, below average
• Prior sweepSwept low
• Open vs London EQAbove EQ
• Four-way34 / 25 / 41 / 0
• First sweep▲ 58% · ▼ 42%
• P(high taken)75% → taken 07:35
• P(low taken)66% → taken 09:30
• Samplen 528, High

Tuesday 25 August

A table the record had not seen, a four-point miss at the high, and the low at 10:20

First sweep ✗ · low only, the 35% branch, two points behind the 37% favourite

A new table. 37 / 35 / 18 / 10 from 215 sessions marked Medium, which the record had not drawn before and would draw again on Thursday: the two exclusive branches two points apart, both sides at only 18%, and an inside session at 10%, the highest that figure has been on any published morning.

Price went at the high first. The session high printed at 29416 shortly after 08:00, four points short of London's 29420, and a second push before 10:00 got no closer. The 10:20 bar then fell through the London low, latched taken at 10:20, and price spent the last hour chopping just beneath it.

Scored, this is a miss on two axes and one from two on the third. The first-sweep call was the high at 53% and the low went; the favourite was high only at 37% and low only at 35% landed. The 53% low marginal resolved taken, the 55% high marginal not. A table with its two leading branches two points apart is a table saying it does not know, and a four-point miss is what that looks like on the chart.

• London range209 pt, above average
• Prior sweepSwept high
• Open vs London EQAbove EQ
• Four-way37 / 35 / 18 / 10
• First sweep▲ 53% · ▼ 47%
• P(high taken)55% → not taken, missed by 4 pt
• P(low taken)53% → taken 10:20
• Samplen 215, Medium

Wednesday 26 August

An 80% call on the high, and the low went first on the 08:30 data bar

First sweep ✗ · both sides, the 50% favourite ✓🔴 Core PCE + GDP · 08:30 ET

The fourth draw of 38 / 11 / 50 / 1, the record's most frequent table, from a 93-point London that stayed inside the prior session. Price opened 35 points under the high with 80% on the high going first, the strongest first-sweep call of the week.

It went the other way. Price sold off from the open and the 08:30 bar, which carried Core PCE and the GDP revision, took the London low. The low of the session printed at 29133 just after 09:00, fifty points beneath the level. The 09:30 bar then reversed two hundred points and took the high, latched at 09:30, and the session faded back into the range by the close.

Both sides at 50% landed, and both marginals resolved taken, the 88% high and the 61% low. The first-sweep call did not: 80% on the high and the low went, on the bar the red folder was marked against. The three earlier draws of this table took the high first every time; this is the first that did not.

• London range93 pt, below average
• Prior sweepInside
• Open vs London EQAbove EQ
• Four-way38 / 11 / 50 / 1
• First sweep▲ 80% · ▼ 20%
• P(high taken)88% → taken 09:30
• P(low taken)61% → taken 08:30
• Samplen 524, High

Thursday 27 August

Tuesday's table again, and Tuesday's outcome: the high missed, the low taken

First sweep ✗ · low only, the 35% branch, two points behind the 37% favourite

37 / 35 / 18 / 10, first sweep 53 / 47, n 215: Tuesday's table, drawn again two days later from a London four points wider. Every week of the record has now produced a within-week repeat, and this week produced two.

The session drifted lower for two hours and took the low at 09:35, with a session low of 29424 twenty-four points beneath it. The recovery ran from there to 29634 at 11:00, 27 points short of the London high, and faded. Low only, again.

The same scores as Tuesday on every axis: the first-sweep call was the high and the low went, the 37% favourite lost to the 35% branch, the 53% low marginal was taken and the 55% high was not. Two draws of a table that splits its two leading branches by two points, and both resolved to the lower one. Two is not a pattern; it is the entry in the record.

• London range213 pt, above average
• Prior sweepSwept high
• Open vs London EQAbove EQ
• Four-way37 / 35 / 18 / 10
• First sweep▲ 53% · ▼ 47%
• P(high taken)55% → not taken, missed by 27 pt
• P(low taken)53% → taken 09:35
• Samplen 215, Medium

Friday 28 August

The tightest London of the record, both sides by 09:30, and a 300-point rally off the 10:00 print

First sweep ✓ · both sides, the 50% favourite ✓🔴 Warsh at Jackson Hole + payrolls revision · 10:00 ET

A 76-point London, the tightest prior session in the four weeks published, and Wednesday's table again: 38 / 11 / 50 / 1 from 524 sessions, its fifth draw. Price opened thirty points under the high with 80% on it going first.

The high went at 08:20. Price held around it for an hour, then fell through the low at 09:30 and kept going to 29505 on the 10:00 bar, the one carrying the Warsh speech from Jackson Hole and the benchmark payrolls revision, seventy points beneath the level. From there the session rallied three hundred points to 29811 by 11:00 and closed near the top.

Three from three: first sweep on the high at 80%, both sides the 50% favourite, both marginals taken. The release fell after both levels had already gone; it marks the session low and the turn, not the take. The week's two draws of this table both resolved both sides, where its first two draws in August had stopped at high only.

• London range76 pt, below average
• Prior sweepSwept low
• Open vs London EQAbove EQ
• Four-way38 / 11 / 50 / 1
• First sweep▲ 80% · ▼ 20%
• P(high taken)88% → taken 08:20
• P(low taken)61% → taken 09:30
• Samplen 524, High

The same forecast, more than once

Across the twenty sessions published so far, seven context bins have each come up more than once, and nineteen of the twenty sessions sit in one of them. Same context bin, same lookup, same numbers to the percentage point, on days weeks apart with different London ranges and different price levels. This week every session was a repeat: Monday, Wednesday and Friday drew tables the first three weeks had already produced, and Tuesday and Thursday drew a table the record had not seen, twice.

Two things follow, and they matter more than this week's score.

The forecast is a lookup, not a fit. Nothing about Thursday the 27th was used to produce Thursday the 27th's numbers. They are the numbers Tuesday got, because both sessions landed in the same measured bin, and the same is true of the three tables this week borrowed from the first three. A table that returns the same answer for the same context bin and then watches the answer resolve both ways is behaving exactly as a table should.

Identical forecasts resolving differently is the system working, not failing. Of the seven bins that have repeated, six have now produced more than one outcome, and the seventh has been drawn only twice. The most frequent table, 38 / 11 / 50 / 1, has now gone high only twice and both sides three times from five draws, against a stated 38 and 50. That is what those two numbers look like at this sample size. Neither settles anything; both are the point.

What this week actually shows

Five things, from a week that scored best on the branch and worst on the first sweep.

1. The both-sides favourite went three from three, which takes it to seven from twelve. Both sides was the highest-rated branch on Monday, Wednesday and Friday, at 41%, 50% and 50%, and all three sessions took both levels. After last week it stood at four from nine across the record; it now stands at seven from twelve. That is one good week added to a mixed record, not a verdict on the branch, and the running tally is the only thing that will become one.
2. A table the record had not seen came up twice and resolved the same way twice, against its favourite. Tuesday and Thursday both drew 37 / 35 / 18 / 10 from 215 sessions, the two leading branches two points apart. Both mornings took the low and stopped, Tuesday missing the high by 4 points and Thursday by 27. Two low-only outcomes from a table that says 37 high, 35 low is exactly as likely as two high-only ones, and the pair is on the page because the record will draw this table again.
3. Two from five on the first sweep, the weakest week on that axis so far. The mean stated share on the called side was 65%, the lowest of the four weeks, because the two 53 / 47 mornings were close to a coin toss and both went the other way. The miss that matters is Wednesday: 80% on the high, and the low went first on the 08:30 data bar. Across the four weeks the first-sweep call stands at twelve from twenty.
4. The per-side view again carried the sessions the other axes buried. Eight of ten levels resolved as their marginals implied, against a mean stated 66%, and thirty of forty across the four weeks against roughly 69%. Wednesday scores nothing on the first sweep and two from two here. If you trade one level at a time rather than betting the branch, that remains your column.
5. Twenty sessions still settle nothing, and the indicator's own row is behind for a fourth week. Forty level events cannot confirm frequencies measured over hundreds. The indicator's realised versus expected row read 50 to 51% against 56% expected across the week, on 262 to 279 setups loaded, which is behind for the fourth week running. The record has a shape, not a verdict, and that row is the number to keep watching.