NY AM, week commencing Monday 31 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 and Friday and above it on the other three days; Tuesday's 379 points is the widest prior session in the record, more than a hundred points beyond the previous widest. Two red folder USD releases fell inside the session this week: ISM Manufacturing at 10:00 ET on Tuesday, two hours after the low had gone, and Non-farm payrolls at 08:30 ET on Friday, an hour in, on the bar that took the low five minutes after the high. Both are marked against the day. Monday 31 August was a normal trading 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 31 August

Opened four points above the London low with 85% on it, and it went in the third bar

First sweep ✓ · low only, the 52% favourite ✓

Price opened four points above the London low with 85% on it going first, and it went in the third bar of the session at 07:40. The table is 10 / 52 / 38 / 0 from 840 sessions, the record's largest sample, drawn on Monday 3 August, when both sides went, and Thursday 20 August, when the low went and nothing else.

This one did what the 20th did. The fall ran eighty points to 29354 by 10:30, the recovery came back to the level itself, and the high, a hundred points above at a 48% marginal, was never in range. Three from three: first sweep, the 52% favourite, and both marginals, the 90% low taken and the 48% high not. A call with the level four points away is closer to arithmetic than to a forecast, as the page said of the 20th.

• London range105 pt, below average
• Prior sweepInside
• Open vs London EQBelow EQ
• Four-way10 / 52 / 38 / 0
• First sweep▲ 15% · ▼ 85%
• P(high taken)48% → not taken
• P(low taken)90% → taken 07:40
• Samplen 840, High

Tuesday 1 September

The widest London of the record, Monday's table, and Monday's outcome

First sweep ✓ · low only, the 52% favourite ✓🔴 ISM Manufacturing PMI · 10:00 ET

London fell 379 points overnight, the widest prior session of the record by more than a hundred points, and the lookup did not change: 10 / 52 / 38 / 0, Monday's table, because the inputs it keys on were the same. Swept low, below EQ, price 35 points above the low with 85% on it going first.

The low went at 07:55, the fall ran to 29040 at 09:45, 137 points beneath the level, and the recovery got back above the low by 11:00. The ISM print at 10:00 landed in the middle of that recovery and did not change its direction. The high, 344 points above the open, was never a question.

Three from three again, and the bin's fourth draw: both sides once, low only three times. On Monday and Tuesday the same forecast met a 105-point London and a 379-point one and resolved the same way, which is the argument the repeat-bins section makes below.

• London range379 pt, above average
• Prior sweepSwept low
• Open vs London EQBelow EQ
• Four-way10 / 52 / 38 / 0
• First sweep▲ 15% · ▼ 85%
• P(high taken)48% → not taken
• P(low taken)90% → taken 07:55
• Samplen 840, High

Wednesday 2 September

Last week's new table on its third draw, and this time the 37% favourite landed

First sweep ✓ · high only, the 37% favourite ✓

37 / 35 / 18 / 10 from 215 sessions for the third time in six sessions. Both earlier draws, Tuesday 25 and Thursday 27 August, took the low, missed the high by 4 and 27 points, and scored against the 37% favourite.

This one rallied from the open, took the high at 08:10, came back to test the level twice, at 09:00 and again at 10:00, and then climbed to 29211 by 11:00. The session low was the opening area at 29017, ninety points above the London low, which was never approached.

Three from three: the high at 53% went first, high only at 37% landed, the 55% high marginal was taken and the 53% low was not. The bin is now one high only and two low only from three draws, which is roughly what 37 / 35 says it should be, and the both-sides branch at 18% has not appeared.

• London range194 pt, above average
• Prior sweepSwept low
• Open vs London EQBelow EQ
• Four-way37 / 35 / 18 / 10
• First sweep▲ 53% · ▼ 47%
• P(high taken)55% → taken 08:10
• P(low taken)53% → not taken
• Samplen 215, Medium

Thursday 3 September

The 25% both-sides branch landed, and then price ran 430 points off the session low

First sweep ✓ · both sides, the 25% branch

The third draw of 24 / 45 / 25 / 6, the table Tuesday 18 August took the low on and Wednesday 19 August took both sides on. Price opened sixty points above the low with 62% on it going first.

The low went at 07:45 with a session low of 29102, four points beneath the level, and that was the bottom. Price reversed, took the high at 09:35, pulled back to it at 10:30, and then ran to 29540 by the close, 250 points above the London high and 430 above the session low.

First sweep correct at 62%. The branch was both sides at 25%, against a 45% low-only favourite, and both marginals resolved taken, the 70% low and the 49% high. The bin is now low only once and both sides twice from three draws, and both of the both-sides mornings were reversals through the entire London range.

• London range188 pt, above average
• Prior sweepSwept high
• Open vs London EQBelow EQ
• Four-way24 / 45 / 25 / 6
• First sweep▲ 38% · ▼ 62%
• P(high taken)49% → taken 09:35
• P(low taken)70% → taken 07:45
• Samplen 406, High

Friday 4 September

A first-appearance table, the high at 08:25, and the low on the payrolls bar five minutes later

First sweep ✓ · both sides, the 24% branch🔴 Non-farm payrolls · 08:30 ET

The only forecast this fortnight the record had not seen: 53 / 20 / 24 / 3 from 479 sessions, with 74% on the high going first. Price opened thirty points under the high on a payrolls morning, with the release an hour into the session.

The high went at 08:25, one bar before the print. The 08:30 bar then fell two hundred points, through the London low, which latched taken on that bar, to 29500. Price rallied all the way back to the London high by 10:00, held under it, and fell again to a session low of 29483 before the close.

First sweep correct at 74%, and both marginals taken, the 77% high and the 44% low. The branch was both sides at 24% against a 53% high-only favourite. The high was taken before the release and the low on it; the calendar was on the page, and that is the honest reading of the morning. First draw of this table, so there is nothing to compare it with yet.

• London range125 pt, below average
• Prior sweepSwept high
• Open vs London EQAbove EQ
• Four-way53 / 20 / 24 / 3
• First sweep▲ 74% · ▼ 26%
• P(high taken)77% → taken 08:25
• P(low taken)44% → taken 08:30
• Samplen 479, High

The same forecast, more than once

Across the twenty-five sessions published so far, seven context bins have each come up more than once, and twenty-three of the twenty-five 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 four sessions were repeats, Monday and Tuesday of each other as well as of two August mornings, and Friday's table was a first appearance.

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

The forecast is a lookup, not a fit. Nothing about Tuesday the 1st was used to produce Tuesday the 1st's numbers. They are the numbers Monday got, and Monday 3 and Thursday 20 August got, because all four sessions landed in the same measured bin. What puts a session in a bin is not how wide London was, which is why a 105-point morning and a 379-point one drew the same table. It is not the prior sweep either: this same cell has now been drawn after a swept high, a swept low and an inside session. 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. All seven bins that have repeated have now produced more than one outcome; the last holdout, 37 / 35 / 18 / 10, went high only on Wednesday after two low-only draws. The largest bin, 10 / 52 / 38 / 0 from 840 sessions, has gone low only three times and both sides once from four draws, against a stated 52 and 38. 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 the best week on the first-sweep axis and an ordinary one on the other two.

1. Five from five on the first sweep, and most of it was arithmetic. The mean stated share on the called side was 72%, and on three of the five mornings the called level was within 35 points of the open: four points on Monday, 35 on Tuesday, 30 on Friday. The two calls that had to do some work, Wednesday at 53% with the high 70 points away and Thursday at 62% with the low 60 points away, both landed too. Across the five weeks the first-sweep call stands at seventeen from twenty-five.
2. The same table met a 105-point London and a 379-point one and resolved the same way. Monday and Tuesday both drew 10 / 52 / 38 / 0 from 840 sessions, because the bin does not key on the width. Both took the low inside the first half hour and never approached the high. Tuesday's London was the widest of the record by more than a hundred points, and the forecast did not know or care.
3. Last week's new table went the other way on its third draw. Tuesday 25 and Thursday 27 August drew 37 / 35 / 18 / 10 and both took the low against the 37% high-only favourite. Wednesday drew it again and took the high. One high only and two low only from a table that says 37 and 35 is about what it should look like, and every one of the seven repeated bins has now split at least once.
4. The per-side view scored seven of ten, and the both-sides favourite did not appear. Seven of ten levels resolved as their marginals implied, against a mean stated 62%, the lowest stated mean of the five weeks; thirty-seven of fifty across the record against roughly 68%, a realised 74%. Both sides was not the highest-rated branch on any morning this week, so its tally stays at seven from twelve. Thursday and Friday landed it at 25% and 24%, which is the branch score's cost and the per-side view's gain.
5. Twenty-five sessions still settle nothing, and the indicator's own row is behind for a fifth week. Fifty level events cannot confirm frequencies measured over hundreds. The indicator's realised versus expected row read 50 to 52% against 56% expected across the week, on 262 to 278 setups loaded, which is behind for the fifth week running. The record has a shape, not a verdict, and that row is the number to keep watching.