Four first sweeps from five, and both misses landed on inflation day

The week at a glance

Every number in this table is transcribed from the close capture of that morning. Nothing is inferred and nothing is computed after the fact.

Four-way reads high only / low only / both / inside and always sums to 100. A ~ marks a cell whose shares moved materially between market eras, so the indicator quotes them as approximate. High and Low carry the marginal probability that side is taken at all, which is that side's exclusive share plus the both-sides share, not the number printed on the level itself. Red folder USD releases are marked against the day; both fell at 08:30 ET, thirty minutes into the session.

Morning by morning

Open capture on the left, close capture on the right. Click any chart to enlarge.

Monday 10 August

The near side went, and nothing else did

First sweep ✓ · modal branch ✓ · low only, as rated

Eurasia swept its own high overnight and then handed US AM a session opening below Eurasia EQ, which puts the low nearer. The tables read that correctly: 87% on the low going first, and it went at 08:20.

After that the session turned and spent three hours climbing back through the middle of Eurasia's range without ever threatening the high, thirty-five points above. Low only at 69% was the favourite and it landed, so the morning scores on both axes.

Axis 3 splits one from two: the low at 82% went, the high at 24% did not. Both resolved the way their marginals said they should.

• Eurasia range35.2 pt, below average
• Prior sweepSwept high
• Open vs Eurasia EQBelow EQ
• Four-way11 / 69 / 13 / 7
• First sweep▲ 13% · ▼ 87%
• P(high taken)24% → not taken
• P(low taken)82% → taken 08:20
• Samplen 562, High

Tuesday 11 August

A mirror of Monday, and the tables turned with it

First sweep ✓ · modal branch ✓ · high only, as rated

The exact inverse of Monday, and worth putting side by side. Eurasia swept its low this time, price opened above EQ, and the whole table flipped: 71% on high only where Monday had 69% on low only, and 87% on the high going first where Monday had 87% on the low.

It went at 08:05, five minutes into the session. Price then spent the rest of the morning well above the low, which sat thirty-six points below and was never in question at 25%.

This is what the tables are: a lookup on the previous session's geometry. Two mornings with opposite setups returned opposite forecasts of near-identical confidence, and both landed.

• Eurasia range36.1 pt, below average
• Prior sweepSwept low
• Open vs Eurasia EQAbove EQ
• Four-way71 / 11 / 14 / 4
• First sweep▲ 87% · ▼ 13%
• P(high taken)85% → taken 08:05
• P(low taken)25% → not taken
• Samplen 603, High

Wednesday 12 August

Both sides on CPI day, one five minutes early and one on the print

First sweep ✓ · modal branch ✗ · both sides, rated ~10%🔴 CPI · 08:30 ET

Core CPI and CPI released at 08:30 ET, thirty minutes into the session. The high went at 08:25, on the drift into the release, and the low went at 08:30, on the print itself. Five minutes apart, in opposite directions.

The tables rated both sides at roughly 10%, the second-smallest of the four branches, and that is the branch that landed. The favourite, high only at about 55%, was half right in the least useful way: the high did go, and go first, but the session did not then leave the low alone.

The first-sweep call survives intact at 72% on the high, which is the axis that asks which side goes rather than how many. It is a fair illustration of why the two axes are scored separately. The four-way branch is a statement about containment, and a release is precisely the thing that removes it.

• Eurasia range40.2 pt, below average
• Prior sweepSwept high
• Open vs Eurasia EQAbove EQ
• Four-way~55 / ~22 / ~10 / ~13
• First sweep▲ 72% · ▼ 28%
• P(high taken)65% → taken 08:25
• P(low taken)32% → taken 08:30
• Samplen 640, High

Thursday 13 August

The week's only double miss, and the release had nothing to do with it

First sweep ✗ · modal branch ✗ · low only, rated 19%🔴 PPI · 08:30 ET

Core PPI and PPI released at 08:30 ET, and gold did approximately nothing with it. The session drifted lower for two hours before the low finally went at 10:30, which is not a release reaction by any reading.

Both axes miss. The tables favoured the high at 48% and on first sweep at 70%. The low went, and it was the only thing that went, on a branch rated 19%. This is the week's one genuinely bad morning and it is worth stating plainly rather than burying: a 70% call resolved the other way.

The other number on this table is the interesting one. Inside was rated 29%, the highest of the week, on the widest Eurasia range so far at 44.9 points. A wide prior session pushes both levels further away and makes containment more likely. It did not happen here, and it did not happen on any of the five mornings.

• Eurasia range44.9 pt, above average
• Prior sweepSwept low
• Open vs Eurasia EQAbove EQ
• Four-way48 / 19 / 4 / 29
• First sweep▲ 70% · ▼ 30%
• P(high taken)52% → not taken
• P(low taken)23% → taken 10:30
• Samplen 573, High

Friday 14 August

The most confident call of the week resolved in the opening bar

First sweep ✓ · modal branch ✓ · high only, as rated

The widest Eurasia range of the week at 60.4 points, well above average, and a session that opened with price already pressed against Eurasia's high. The tables put 92% on the high going first, the most confident call of the week, and 76% on high only.

It went at 08:00, in the opening bar. Price then held above it for the rest of the morning, and the low, sixty points below at 12%, was never a live question.

An easy morning, and worth being honest about why it was easy: when a level is already within touching distance at the open, a high stated probability is close to arithmetic. The mornings that test the tables are the ones like Thursday, where both levels are in play and the call has to choose.

• Eurasia range60.4 pt, above average
• Prior sweepSwept high
• Open vs Eurasia EQAbove EQ
• Four-way76 / 7 / 5 / 12
• First sweep▲ 92% · ▼ 8%
• P(high taken)81% → taken 08:00
• P(low taken)12% → not taken
• Samplen 634, High

What the week actually showed

• Both modal misses fell on the two inflation mornings, and that is an observation, not a finding. Wednesday carried CPI and Thursday carried PPI, both at 08:30 ET, and they are the only two mornings the four-way branch missed. Two events cannot establish anything and we are not going to pretend otherwise. It is the reason the calendar is now marked on every one of these pages: if the pattern is real it will show up over a year of them, and if it is noise that will show up too.
• A release breaks containment, not direction. On Wednesday the first-sweep call was right and the four-way branch was wrong, which is the signature you would expect. The four-way branch is a claim about how much of the range gets used, and a release is the thing that removes the limit. The direction call, which only asks which side goes first, came through Wednesday at 72%, and failed on Thursday for reasons that had nothing to do with the print two hours earlier.
• Nothing stayed inside, on forecasts that expected it up to 29% of the time. Across five mornings the tables averaged roughly 13% on the inside branch and it landed zero times. At this sample size that is unremarkable in both directions, and it is worth saying so before anyone reads a pattern into it.
• Levels went slightly more often than stated, by an amount that means nothing yet. Six of ten against a mean stated 48% is about three quarters of one standard error on ten events. It is in the table because it belongs in the table, not because it is evidence.
• The most confident call was also the easiest. Friday's 92% first sweep resolved in the opening bar, because the level was already within touching distance when the session opened. High stated probabilities on near levels are closer to arithmetic than to insight. Thursday, where both levels were live and the call had to choose, is the one that tested the tables, and it is the one they got wrong.