Every instrument in this morning's scan sits inside its own entry window and not one has crossed into it. Zero qualified, against three yesterday. Overnight the stocks leg switched off, so a cycle that had bonds, stocks and risk appetite behind it a week ago now runs on risk appetite alone. Astera Labs and CECO are the two setups, and Astera's cost line is the number that decides the page.
Futures pricing put the odds of a Federal Reserve hike this month at 66% at Tuesday's close, roughly double the 35% priced on 27 August, before Kevin Warsh told Jackson Hole the Fed has work to do unless it can be confident underlying inflation is heading to target. Michael Barr followed on Tuesday, saying the Fed should act decisively to raise rates if inflation is not moderating. The ten-year went to 4.79%, its highest in twenty months, and the thirty-year near 5.27% is on its worst run since 2006. A market that opened the year pricing cuts now prices a hike in fourteen days. That is not a headline risk. It is the discount rate underneath every long on this page, and it is why the bond leg of the framework has been off since yesterday morning.
Yesterday the money-flowing leg went out. Overnight the stocks leg followed it, and that is the whole difference between an inflationary pump and a speculative one: the cycle is now carried by risk appetite on its own, the tailwind that paid longs all last week has drained from +5 to +0, and the strength score holds at +3 only because of what is left. Nothing on this page takes a size from the regime this morning. It takes a size from its own light or it takes none. One structural caveat, the same feed split as yesterday and wider than yesterday. The six US-exchange charts print the reading in the bar above. The four index CFDs print full risk-off at +0 of 10.5, all three legs down, on a −2 tailwind. The exchange feeds govern, as they have all week, and the gap between them is a symbol-resolution artefact rather than two opinions.
The strip runs when it argues with the regime, and this morning it argues with itself. The banner reads breadth recovery. Inside the same 74-name universe, advancers went from 44 to 20 in a single session, and the summation index turned down for the first time in the sequence.
| Reading | Tuesday close | Monday close | Direction |
|---|---|---|---|
| Advancing versus declining | −34 net, 20 advancing | +14 net, 44 advancing | Collapsed |
| Advance-decline ratio | 0.37 | 1.47 | Weaker |
| Advance-decline line | +988 | +1,102 | Weaker |
| Above the 50-day line | 27.0% | 59.5% | −32.5 points |
| Above the 200-day line | 35.1% | 68.9% | −33.8 points |
| McClellan oscillator | −51.2 | +25.9 | Turned negative |
| Summation index | +11,499 | +11,628 | First fall |
Seven columns, seven moves the wrong way, and five of them are not small. The fifty-day column gave up 32.5 points and the two-hundred-day 33.8, so the columns say roughly a third of the universe fell through both of its own moving averages inside one session. The McClellan went from +25.9 to −51.2. The summation index, which had climbed through three straight sessions of narrowing, finally gave up 129. The banner is measuring breadth against a price that fell 0.71%, and on that narrow test it is right. On every test that decides a position it is wrong, and that is why three of the four sized rows below carry a half or a hold rather than a full.
Four names out of ten carry a size this morning, one at full and three at half or hold, against seven of sixteen yesterday. The change that matters is that not one instrument qualified. All ten windows in the scan are open and all ten are still waiting on the cross, which has not happened on a single session since last Tuesday. Every trigger below is a four-hour close in the trade's direction, never a touch, and the get-out is always a close back through the cost line. The other six sit in Watch and Avoid, and four of those are the indices.
| # | Name | Side | Entry · trigger · get out | What the terminal reads | Size |
|---|---|---|---|---|---|
| 1 | ALAB Astera Labs | Short | 🔒 Members | The only fully wound coil in the scan and the highest confirmation on the page, with the separate institutional cost lines clustered tighter than anywhere else here. The record is the problem, and it is the whole reason this is a half.Record not published, 6 resolved of 20 needed, 7 fired, 1 open · window open, not yet qualified · eight readings across two timeframes · compression building at 100%, the fullest reading here · convergence active at 90% | Half |
| 2 | CECO CECO Environmental | Short | 🔒 Members | Best publishable record in the brief and the only green light attached to one. It read long here yesterday and that published long was closed by its own get-out overnight. The direction has now turned to meet the volume, which never confirmed the long in the first place.Record +.17R on 120 signals, the highest here · window open, not yet qualified · six readings across three timeframes · compression building at 95% · sellers dominant by 22% · light GREEN, fading | Full |
| 3 | TM Toyota Motor | Long | 🔒 Members | The reclaim called in the members chat is onside and the light is still green. The overnight print came in just above the price that ends it, which is close enough to make this a position you hold rather than a row you add to.Record +.15R on 118 signals · window open, not yet qualified · six readings across three timeframes · sellers dominant by 17% · light GREEN, holding, 10.1-point cushion | Hold |
| 4 | CRCL Circle Internet Group | Short | 🔒 Members | Convergence running at 76% with nothing wound behind it, on the youngest sample in the scan. Half is the ceiling here until the record exists, however the chart looks.Record not published, 4 resolved of 20 needed, 4 fired, 0 open · window open, not yet qualified · six readings across two timeframes · convergence active at 76% · compression expanded at 50% | Half |
If you watch one, watch Astera Labs. CECO holds the better record and the safer size, and on a normal morning CECO is the row you take. But Astera is the only instrument in the scan with its range fully wound, the highest confirmation on the page, and the tightest cluster of institutional cost lines behind it. On a session where every window is open and empty, a coil at 100% is the only stored move on this page.
Enter your Cantillon Research subscriber email to unlock the complete September 2, 2026 brief: all four sized rows with their windows, triggers and get-outs, the cost line underneath Astera’s coil and the two closes that settle it, CECO’s level after yesterday’s published long was stopped out, Toyota’s overnight print sitting just above the price that ends the position, the six instruments in Watch and Avoid, and all ten terminal captures.
Cost line (the anchor) · the volume-weighted price large money actually paid over the current swing. Everything in this brief is measured as distance from it, and that distance is what "deviation" means. Sell near it and you are selling where they bought. Sell far below it and you are handing them the profit. Entry window · how far from the cost line a new entry can still be sensibly priced. The width comes from the instrument's own volatility tier: two percent for indices and large caps, three percent for mid-range names, three to five for the high-volatility list. Open means you are inside it and waiting on the cross. Qualified means the cross has already printed inside it. Shut means the move has run and you wait for a pullback or a fresh coil. Every trigger in this brief is a four-hour close, in the trade's direction, never a touch. Compression · how tightly the range has wound, from 0 to 100%. Building means it is loading. Coiled means it is nearly full, and coiled spring is the terminal's own label for the tightest version. High compression means the next move should be a big one, says nothing about direction, and never overrides a shut window or a red light. Confirmation · how many independent readings agree, and across how many timeframes. Higher on both counts is better, and two timeframes at 6 beats one timeframe at 8. Buyer or seller dominance · which side is actually transacting at the most-traded price, as a percentage skew. Neutral means the two sides are matched and the volume is telling you nothing about direction. Snapback reset · the terminal's oversold or overbought extreme, where price pushed through the second deviation band and reclaimed it. The bar count since the last one is how fresh the current structure is. Stability · a two-part check that the current reading has held long enough to be relied on. Zero of two means neither part has passed, which is every instrument on the page today. Convergence · how tightly the separate institutional cost lines have clustered. High convergence means several cohorts of large money paid roughly the same price, so the line underneath is thicker than usual. Record · the average result per signal in units of risk, where being stopped out is exactly −1.00. Always published with the number of finished signals behind it, and never published at all below 20 of them. It is in-sample, so read it as a description of the past and not a forecast. Where a record field is distorted or faulty, this brief says so and does not print the number. Trend light and cushion · green means full size, amber means half, red means no new entry however good the chart looks, and unproven means the sample is still being built. The cushion is how far the light is from changing, and a negative cushion means it has already changed against you.
Readings captured Wednesday 2 September 2026, 06:28–06:46 CEST, from the IVT v14 terminal, on four-hour bars that had not yet closed. That is why every trigger here is a close and not a touch, and this morning it decides two of the four rows outright: Toyota’s overnight print sits above the price that ends the position and Astera’s sits below the price that starts one, and neither of those prints is a close.
Not one instrument in the scan qualified. Every window is open and every cross is still pending, so every entry on this page is conditional on a bar that has not printed yet.
The four index captures run on broker cash CFD feeds and resolve the regime panel differently from the six US-exchange captures, printing full risk-off at +0 of 10.5 against +3 on the exchange charts. The exchange reading governs throughout this brief, as it has all week, and the difference is a symbol-resolution artefact rather than two regimes. Index levels will also differ from your own index or futures contract.
Records are in-sample, calculated over the same history the settings were chosen on, so they describe the past and forecast nothing. No rate is published on fewer than twenty finished signals and the sample size is printed beside every one, which is why Astera Labs and Circle carry raw counts instead of numbers. Entry windows are derived from the percentage distances the terminal publishes, so treat them as accurate to the nearest cent or index point; trigger levels are set at the nearest structural shelf and are ours, not the terminal’s.
The Toyota position referenced in this brief was called in the members chat on the reclaim, not in the published board, where Toyota was a half-size short closed by its own get-out on Monday. The Russell reference in Watch and Avoid is a portfolio close, not a published board row. Both are stated so the record reads straight.
Breadth columns compare Tuesday’s close against Monday’s across the same 74-name scan universe. The advance-decline line and summation index are cumulative series and are not comparable to the single-session columns beside them.
Charts are unmodified terminal captures provided for reference, not as entry instructions. Educational content only. Not financial advice. Past results do not predict future ones.