The cycle turned. My largest position is the one it punishes.

Where we are

The business cycle confirmed SUMMER this week, ending five consecutive weeks in FALL. SUMMER is the inflationary boom quadrant: growth accelerating while inflation rises. The engine marks it confirmed on six consecutive daily reads with high confidence and no provisional reading pending, and for the first time in five editions the quadrant is not sitting on its own boundary. Its asset tilts are unambiguous, and they are why this edition is about a portfolio rather than a chart — tailwind for value, energy, materials and financials; headwind for duration; tailwind for commodities; and headwind for cash, which SUMMER erodes by construction.

The quadrant did not change because inflation cooled. It changed because growth firmed while inflation kept rising, and the inflation composite is now unanimous for the first time in this window: core PCE at 3.29% year over year, core CPI at 2.79%, and five-year breakevens at 2.34%, all reading rising. The August 17 edition noted that only breakevens were stable. They are not any more. Headline CPI is 3.54%, headline PCE 3.67%, average hourly earnings 3.15% and rising. Inflation remains the dominant risk factor. Growth strong with inflation rising is a different environment from growth weak with inflation rising, but it is not a calmer one. It is the quadrant where central banks tighten into strength.

The market regime is unchanged and unresolved: YELLOW since August 3, with consumer sentiment at 55.2 against a RED/YELLOW threshold of 55.0 — a margin of 0.2, for the third consecutive week, on a print dated July 31. The feed publishes its final number at month-end and the collector returned no new figure, which is the feed behaving normally rather than a gap. The next final print is due Friday, August 28. A colour change from here is a survey revision crossing a line, and it would be as likely to go RED as GREEN.

The evidence that tipped it is a month old

The growth composite is what moved, and it is exactly what the last edition named as standing between FALL and SUMMER. That edition had two of three components positive with the manufacturing proxy neutral. This week all three are positive: manufacturing proxy, initial claims at a 204,000 four-week average, and consumer sentiment. One component flipped and the quadrant followed.

The caveat belongs with the number rather than after it. Both newly supportive growth inputs are July observations. The manufacturing proxy is industrial production, and the observation that moved it is dated July 1 — released August 19, which is why it lands in this week's read, but it describes July. Consumer sentiment is the July 31 print, now 24 days old. Only initial claims, current to August 15, is genuinely recent. The cycle turn is real and confirmed by the engine's own rules. The picture it confirms is a July picture.

And the July 31 sentiment print is doing double duty, which is the sharpest thing in this edition. The same number sits 0.2 from the RED/YELLOW threshold as a level, and scores as one of three positive growth components as a direction. Those are two different measures of one survey, and Friday's revision would touch both — not necessarily in the same direction or by the same amount, but not independently either. A reader treating the cycle turn and the softened regime colour as two separate pieces of evidence is counting one survey twice.

Some structural context on how much weight a fresh quadrant flip can bear. Over 17 weeks the quadrant has changed four times: SUMMER to FALL on June 1, back on June 29, back again on July 20, and to SUMMER this week. The regime colour changed twice more. This SUMMER call is six days old against the four weeks the FALL call had accumulated before it broke. Half the history — nine of eighteen weekly rows — is reconstructed point-in-time rather than recorded live, so revisions are not reflected at the early end. And the oldest input under the headline is 54 days old: the Chinese central bank balance sheet, carrying a July 1 observation date while the composite carries today's timestamp.

The plumbing disagreed with its own label

The last two editions built toward a specific test: three consecutive weeks of global liquidity expansion would put the dominant force of this quarter genuinely in question. Week three did not clear it. On a constant three-bank basis — Fed, ECB and Bank of Japan at both ends, so coverage changes cannot fake a move — liquidity is $17.66T, down $40B on the week. It rose $220B into August 10 and $60B into August 17, then gave $40B back. The trough remains August 3 at $17.42T, the three-week net is $240B, and the twelve-week trend is still a 1.18% contraction.

The composition is more informative than the size, and it settles a question the last edition left open. The Fed shed $10B to $6.75T, the ECB added $10B to $6.86T, and the Bank of Japan gave back $40B to $4.05T. Tokyo was the entire move, and it is the same balance sheet that supplied most of the two-week rebound. Last week we warned that the rebound might be a weaker dollar revaluing non-dollar balance sheets rather than policy. This week runs the other way and is therefore evidence: the broad dollar fell again, by 0.2%, which mechanically raises the dollar value of ECB and BoJ holdings — and the BoJ dollar figure fell anyway. That is a balance sheet shrinking against a currency tailwind, not a revaluation artefact. Including China, unchanged on the July statement at $7.45T, the four-bank aggregate is $25.10T against $25.14T.

Financial conditions are where this edition has to be careful, because the composite reads loose with direction easing, and this week's pillars do not support that pairing. High-yield spreads widened 4 basis points to 275. Investment grade widened 3 to 82, the widest in the twelve-week window. Volatility rose 1.4 points to 16.0 from last week's window low. Only two pillars improved and both barely: the 10-year real yield eased 1 basis point to 2.35%, still restrictive and still 28 basis points higher over twelve weeks, and the dollar slipped 0.2%. Four of five pillar readings moved toward tighter and the composite direction printed easing.

That is worth stating plainly rather than repeating the label: this week the easing print is a summary that its own components did not earn. It does not make the composite wrong — the pillars are still loose in level, which is what the composite measures — but anyone reading the direction as fresh evidence of loosening is reading more than the week contains. A related discrepancy is disclosed at the end of this edition.

Running the framework on my own book

Jawz maintains a model portfolio in public. It started on May 6 at $1,000,190 and is worth $1,012,010 today. Every position ships with a written thesis and pre-stated falsifiers, and the record is public because a framework that only reports its good weeks is marketing.

Feeding the five positions through the portfolio-wide drift sweep returns exactly one flag, and it is the largest position in the book. SGOV, the Treasury bill sleeve, at 38.6% of net asset value: regime fit 2 out of 5, unchanged since entry, classified structurally low rather than deteriorating — opened low-fit by design, and SUMMER does not favour cash.

The flag is mild. The review question attached to it is not. It quotes the position's own thesis back at it: hold dry powder while real yields are restrictive and get paid to wait for a regime that rewards risk — is that exact claim still supported by current evidence?

The honest answer is that it is weaker than it was, and weaker in a way this edition has already qualified twice. The growth evidence that turned the quadrant is July-dated. The conditions label that would corroborate it printed easing on a week when four of five pillars moved tighter. Real yields are still classified restrictive, which is the exact condition the thesis named for holding. So the thesis is not falsified. What has changed is that it is now being defended by a July picture and a label its own components did not earn, rather than by the plain reading it had a month ago. Nearly two-fifths of this book sits in the single asset class the current quadrant explicitly punishes, on a rationale that has stopped being comfortable. That is the most uncomfortable sentence in this edition and it is about my own portfolio, which is the only reason it is worth reading.

What it has cost: 462 basis points

Since inception on May 6 the book returned 1.18% over 110 days. The S&P 500 over the identical window returned 5.80%. The regime-aware portfolio is behind buy-and-hold by 462 basis points in its first 110 days, and the reason is almost entirely the cash.

Two caveats, both real, neither of which rescues the number. Distributions are credited on neither side, so the comparison is like for like, but this book yields materially more than the index — crediting them narrows the gap by roughly 50 to 60 basis points in our favour and does not close a 462 basis point hole. And 110 days cannot judge a mandate whose primary target is measured trough to trough across a full regime cycle, and whose claimed edge is avoiding drawdowns rather than capturing rallies. There has been no drawdown to avoid.

The honest reading is that this is what the insurance costs, priced in basis points, before anyone knows whether it pays. The mandate explicitly traded upside for halved drawdowns. 462 basis points is the premium paid so far. Publishing it while the outcome is genuinely unknown is the entire point of running a book in public rather than describing one afterwards.

One thing I cannot tell you and will not estimate: whether the book clears its own defence floor of beating Treasury bills over a rolling year. Answering that needs distribution accounting the vault does not have. It has been an open engineering item since August 5 and this is the second review it has compromised, so it is now blocking rather than deferred. A number I cannot produce is recorded as unanswered, not filled in with a plausible one.

The trigger I missed

On August 5 a rotation was armed in writing: sell a slice of the quality-equity sleeve, buy energy equity, execute when the energy fund holds above 58.52 on two consecutive reads with the regime YELLOW or better and the cycle confirmed. Falsifiers were written the same day.

It read 61.03 on August 13, 61.91 on August 18, 63.64 on August 22 and 63.11 today. The daily health check saw it and escalated. It escalated again on the 18th, in writing, noting the trigger had been sitting at its level since the 13th. Both escalations were correct. Both were ignored, because the weekly portfolio review that owns this trigger was skipped three weeks running while attention went to other work.

I executed it today. Entry is 1.9% worse than the earliest date the conditions were provably met, about $980 on a $50,600 position. The failure was not in the model — the model fired on time, twice, in writing. The failure is that an armed trigger nobody actually reviews is not a system, it is a note. The fix is a rule rather than an intention: an escalation that repeats twice without a decision recorded either way becomes blocking. A trigger may be declined. It may not be ignored.

The trade needed re-testing rather than assuming, because it was written for FALL and FALL ended this week. Its stated falsifier was the inflation composite flipping to tailwind. That has not happened — inflation is still classified headwind and now unanimous in rising. The quadrant changed label because growth improved, and SUMMER names energy explicitly among its favoured sleeves, so the premise survives the relabelling. The one thing the delay genuinely broke was the risk parameter: holding the original abort level from a higher entry would have silently doubled the risk budget, so the abort moved up to the level that defined the original observation.

Why the cash stays anyway

Having said the cash is harder to defend than it was, I did not cut it. That needs an argument, not a shrug.

The mandate sets a risk ladder in advance: YELLOW permits 55 to 65% in risk assets, GREEN permits 75 to 85%. The book sits at 61.3%, already in the upper half of the YELLOW band. Cutting the bill sleeve meaningfully would push through 65% — taking GREEN-regime risk on a YELLOW print two-tenths of a point from RED, in the same week a whipsawing quadrant flipped on a six-day streak, on July data, with a conditions label its own pillars did not earn.

So the drag is not an oversight. It is the mandate executing as designed. This book is behind the index because a rule written in advance caps its risk, and that rule has not yet been paid for by the drawdown it exists to avoid. Overriding it now — after 110 days of underperformance, on the day the cycle turned favourable, with the framework itself flagging the position — would be the most predictable way imaginable to destroy the thing being tested. Frameworks are not worth much on the weeks they agree with you.

What would change it is written down and needs no fresh thinking when it arrives: a confirmed GREEN print moves the band to 75 to 85% and the cash cut follows mechanically. With sentiment at 55.2, RED is as near as GREEN. Both directions stay live, and Friday decides.

Where this gets written down

Everything above exists as a record because of a deliberate habit, and the mechanics are worth showing rather than the discipline being asserted.

The rotation began as a proposal on August 5 — a file stating the trade, the arming price of 58.52, the two execution conditions, two abort levels, and the falsifiers that would end it once entered. It became a decision record today, stating what was executed, at what price, how late, and why the premise survived a regime change it was not written for. Alongside it sits a composition review holding the performance number, the cash argument and the ladder constraint that produced it. Each file references the others and the mandate they descend from.

The point is not tidiness. Without the August 5 proposal naming a price and a date, this week's miss would be a feeling — a vague sense of having been slow. With it, the miss is 1.9% and eleven days, and the escalations that went unanswered are timestamped. A written falsifier is what makes it possible to be caught, including by yourself, and especially in hindsight, when the temptation is to remember having thought something more sensible than you did.

That is the workflow Rumo exists to hold: proposals before outcomes, decisions that reference the proposal that produced them, reviews that reference both, and falsifiers written while the answer is still unknown. Jawz reads the world; the record is what makes reading it accountable. The 462 basis points above are only meaningful because the reasoning that produced them was written down before the number was.

What to watch

• Friday, August 28. The month-end sentiment print settles the 0.2 margin that has been open for three editions, and it settles it for two separate things at once — the headline colour, and one of the three growth components that turned the quadrant. It is also the PCE release. One survey revision moves more of this read than anything else on the calendar.
• Whether SUMMER holds a second week. Four quadrant changes in seventeen weeks is the base rate, and this call is six days old on July-dated evidence. A second confirming week makes it a cycle read; another flip makes the indicator the story rather than the economy.
• The inflation composite, now load-bearing for two separate things: it is the written falsifier on the new energy position, and it is what keeps the quadrant in SUMMER rather than SPRING. All three components read rising. A flip to tailwind ends the energy sleeve by its own rule.
• Whether the conditions composite prints easing again on a week its pillars support. This week four of five moved tighter. One more divergence of that shape would make the composite direction a methodology question rather than a market observation, and we would say so.
• Global liquidity, where the three-week bar was set in advance and missed. The Bank of Japan gave back $40B against a currency tailwind, which is a real contraction rather than a revaluation. The Chinese leg is frozen until the mid-September statement, so the next three weeks of this series are a G3 story.
• Data quality is clean: all 27 sources current, nothing stale, the daily collector at 26 successes and zero failures. Two defects disclosed rather than hidden. The event-consensus table is unreachable, so the August 28 PCE, September 4 payrolls and September 10 CPI carry no consensus — where a consensus is missing the honest move is qualitative scenarios, not an invented number. And the August 17 edition described that week's conditions direction as easing; the stored row for August 17 now reads stable. Every other figure in that edition matches its stored row exactly. The cause is not yet diagnosed, this edition does not claim to know it, and if it resolves as an error in a published edition it will be corrected in public with the reason.

The frame, not the trade

The specific way to be wrong this week is to treat a quadrant flip as an instruction. SUMMER arrived, it favours energy and commodities and punishes cash, and there is a version of this edition where I read that, cut the bills, chased the sleeve that had already run past its arming price, and called it discipline. Three separate things argued against it: the evidence was July-dated, the corroborating label was not earned by its own components, and a rule written before the week happened said the risk band was already full. The framework said one thing and the mandate said another. The mandate is the one that was written before the week.

The question that puts to a real book is uncomfortable in the right way. Find your largest position and ask what the current regime does to it — not what you believe about it, what the quadrant does to it. Then find the sentence you wrote when you bought it and check whether the condition it named has already been met. Mine had, on evidence I then had to discount twice. Then ask the one that actually matters: when the analysis says trim and a rule you set in advance says hold, which do you obey today? If the answer is whichever feels right this week, you do not have a framework. You have a mood with a spreadsheet attached.

None of this is advice and none of it is a forecast. It is a book, its reasoning, and its cost so far, published while the outcome is still unknown. Come back when the drawdown arrives and we will find out together whether the 462 basis points bought anything.