Liquiditrax
Take the quiz
Liquiditrax Command & Research

Market Regime

Macro regime
Live note
Note updated

Our current view

State (current)

Monday answered the question we set on 17 August about WHERE the Fed's hold gets expressed, and the answer was emphatic. The 30-year Treasury yield rose to 5.31%, the highest since June 2007, on a day when the September hike probability FELL to roughly 30-31% from about 33%. Those two facts move in opposite directions and both are true, which is the single most important thing in the bank this morning. The front end is pricing a Fed that is done, while the long end is pricing something the Fed does not control: oil, and government borrowing. The cited drivers for the 30-year were persistent-inflation worry and government borrowing, against a July budget deficit of $432B versus $346B projected and a 30-year auctioned at 5.216% only four sessions earlier. The curve is not disagreeing with itself by accident. It is separating the policy question from the fiscal-and-energy question, and only the first one is priced as resolved.

The trigger was geopolitical and dated. The 60-day US-Iran peace deadline expired Monday with no compromise, the memorandum of understanding between the two lapsed, Iran ruled out direct negotiations, and Brent hit $90 a barrel after Trump said he does not see the war ending anytime soon. Equities took it as a risk event rather than a growth event: S&P 500 -0.52% to 7,745.06, Dow -0.51% to 53,459.78, Nasdaq -0.32% to 26,644.91, with the weakness arriving in afternoon trade as oil rose. That is the third of the three invalidation legs we wrote on 16 August firing, an oil move that reaches the portfolio through the pump and through term premium simultaneously, and it fired before the Fed minutes rather than after.

Underneath, the US data run split rather than softened. Empire State manufacturing rose 5 points to 20.6, its highest in more than four years, with new orders 17.3 and PRICES PAID up 6 points to 58.6. That is a hot regional print on both activity and cost, landing inside a data run that has otherwise been soft on retail sales, CPI, PPI and sentiment. One regional survey does not overturn a national consumer read, but it does mean the "weakness hold" story that the retail sales miss started is not yet the only story available, and Wednesday's minutes now land on a committee whose three dissenters argued precisely that supply shocks plus the AI capex boom could entrench inflation.

The global picture weakened materially and in a way that supports the demand-destruction reading. Japan's Q2 GDP came in at +0.3% q/q and 1.1% annualised against consensus of +0.5% and 2.0%, with PRIVATE CONSUMPTION at -0.0%, its first negative quarter in eight, capex -1.2%, and the entire growth contribution coming from external demand at +0.5pp while domestic demand subtracted 0.2pp. China's July activity data missed on all three lines: industrial production 4.5% y/y against 4.8% expected, retail sales 0.6% against 1.5% expected, and fixed asset investment -6.7% year to date against -6.0% expected and -5.7% at the half. Canada's July CPI accelerated to 3.0% y/y on gasoline at +25.7%, with ex-gasoline CPI at 2.2% for a third straight month and core measures at 1.9%-2.0%. The world's second and third largest economies are slowing, and the one G10 CPI that printed showed the energy shock sitting entirely in the headline and not reaching core.

Timeline

Our calls (journaled)

  • 2026-08-03: Regime = risk-on while 10Y contained. Invalidation: 10Y breakout > ~4.9%. Outcome: open (10Y at 4.65% on 08-08, well contained; call holding)

  • 2026-08-08: The jobs-driven yield drop is the fuel; gold at a record under the tape is the hedge. Watch whether the 10Y stays under ~4.9% and whether gold keeps a bid on up days. Outcome: open (holding into 10-08; gold still bid at $4,400, VIX sub-15)

  • 2026-08-10: With hike odds already at 40%, Wednesday's CPI is now the regime's swing variable, not the jobs number. The tape has effectively pre-traded a friendly print. Invalidation for the calm: a hot CPI that lifts the 10Y toward the high-4s. Outcome: open (10Y rose to 4.71% on 08-10 on oil, moving toward invalidation before the print; call holding but tightening)

  • 2026-08-11: The invalidation variable is now moving on its own. The 10Y drifting to 4.71% on oil, before a CPI the tape has pre-traded as friendly, means the risk is two-sided: energy-led inflation can lift yields even if core CPI behaves. Watch the 10Y into and out of Wednesday, not just the CPI headline. Invalidation unchanged: a decisive move toward the high-4s. Outcome: partly validated 08-12 - the call to watch the 10Y over the headline was right: oil jumped ~$4 on Iran but the 10Y eased to 4.69%, so the bond market did not treat the energy move as durable inflation. Yield discipline intact into the print.

  • 2026-08-12: Pre-CPI, the cleanest single read is the 10Y's reaction to the print, not the y/y headline. With oil re-adding a premium yet the 10Y easing to 4.69%, the bond market has room to fade a soft print (already priced) and to punish a hot core hard. Invalidation for the risk-on regime unchanged: a decisive 10Y move toward the high-4s (~4.9%). Base case holds risk-on while the 10Y stays contained; the swing is the CPI + how yields absorb it. Outcome: VALIDATED 08-13 - the "watch the 10Y, not the headline" read was correct. CPI printed in line/soft and the 10Y did not move (4.69%, +0.004pp), confirming the friendly print was pre-traded; equities edged up rather than ripped. The regime stayed risk-on with the 10Y contained, exactly as framed. First call to close as a hit.

  • 2026-08-13: CPI is resolved and the 10Y absorbed it without moving, so the regime read shifts from "pending on a swing variable" to "confirmed risk-on while the 10Y stays contained." The remaining tail is oil (WTI ~$83 on Iran) feeding into PPI Thu; that is the one path back to yield pressure. Invalidation unchanged: a decisive 10Y move toward the high-4s (~4.9%), now most likely to come via energy rather than core CPI. Watch PPI and whether oil's premium sticks. Outcome: open

  • 2026-08-14: The regime is no longer waiting on anything binary. Both inflation gates cleared, the 10Y fell rather than merely holding, and the oil tail that was the stated path back to yield pressure deflated on demand rather than diplomacy. That is the cleanest version of the risk-on setup we have had since 03-08. The discipline now is to notice what was NOT resolved: core PPI ex food, energy and trade services ran +0.4% m/m, so the headline disinflation is energy-led and core PCE pass-through is still open. Invalidation unchanged in level (decisive 10Y move toward ~4.9%) but the likely trigger has rotated: with oil soft, the more plausible path is a sticky core print or a hot retail sales/consumer read reviving the hike debate, not energy. Watch retail sales and UMich Friday. Outcome: PARTIAL MISS 08-15 on direction, HIT on the variable. We named retail sales and UMich as the next thing that mattered and they were exactly what moved the tape, so the variable was right. The direction was wrong in an instructive way: we framed the consumer as an inflation risk (a hot read reviving the hike debate) and it arrived as a growth risk instead, retail sales -0.6% and sentiment 51.0. The regime was not invalidated, the 10Y stayed in the 4.6%-4.7% area and hike odds fell to ~32-33%, but our stated risk was one-sided when the data was two-sided.

  • 2026-08-15: The regime's risk has rotated from inflation to demand, and the two are now pointing opposite ways in the same release. UMich showed households spending less while expecting MORE inflation (4.3% from 4.2%), which is a stagflation-shaped print, not a disinflation one. Meanwhile the front end prices a hold and the long end priced a 30Y auction at 5.216%, the highest in 25 years, so the curve is telling two different stories about the same economy. For an allocator that means the single-variable framing we have used since 03-08 (watch the 10Y) is no longer sufficient on its own: watch the SHAPE, front end versus long end, because a hold that comes from weakness is not the same regime as a hold that comes from disinflation. Base case stays risk-on, and Friday's -0.17% with VIX falling to 14.25 and small caps up was rotation, not risk-off. Invalidation now has two legs: the original decisive 10Y move toward ~4.9%, OR a second consecutive weak consumer read that turns the equity market's growth interpretation into an earnings interpretation. Watch Fed minutes and whether the retail miss is revised or repeated. Outcome: open

  • 2026-08-16: Two calls, one on structure and one on the week. On structure: with the first hike priced fully into 2027 and only 23bp of tightening priced by year-end, the "watch the 10Y" framing that served from 03-08 to 13-08 has done its job and is now low-information on its own. The regime's real question has moved to whether the hold is a disinflation hold or a weakness hold, and the curve is currently arguing both, front end pricing a benign pause while the 30Y auctions at 5.216% on deficits and energy. Our new primary read is the SHAPE plus the consumer, not the level of the 10Y. On the week: this is the first week since 03-08 where a macro thesis gets tested by company accounts rather than by another survey. Retail sales and UMich are self-reported and revisable; Walmart's $186.8B and Home Depot's comp are audited. If the retail block comes in materially below the season's 87% EPS beat rate, the growth interpretation of Friday earns a second, harder source of evidence and the equity market's rotation becomes a re-rating. If the block holds, Friday's print looks like noise and the risk-on regime gets extended without a hike to fear. Third leg: the Hormuz escalation now reaches the portfolio through the consumer, not just through yields, because $4.07 gasoline is both an inflation-expectations input and a discretionary-spend drain on exactly the companies reporting. Invalidation for risk-on now has three legs: the original decisive 10Y move toward ~4.9%; a second consecutive weak consumer read, which this week can now deliver in company form; OR an oil disruption that lifts gasoline enough to move UMich inflation expectations again while spending falls. Watch FOMC minutes Thursday for how much of the committee shares the market's 2027 view, and watch flash PMIs Friday against the 54.5 July composite. Outcome: open

  • 2026-08-17: Two additions, one correction of emphasis. The correction: we have been describing the regime as "the market priced the first hike into 2027, so the rate question is closed." Monday's detail says that is a market fact, not a Fed fact. Three regional presidents voted to hike six weeks ago, the Chair has removed forward guidance so every meeting is live, and the committee's own hawks argued that supply shocks plus the AI capex boom could entrench inflation, which is a structural argument that soft CPI prints do not answer. The rate question is priced, not settled, and Wednesday's minutes are the cheapest available test of the gap between those two things. The addition: the second-order read this week is that the Fed's hold is being delivered through the long end and financial conditions rather than the policy rate, which is why the 30-year at 5.216% and the steepening are the stance, not a sideshow. An allocator holding duration should understand they are exposed to the instrument the Fed is currently leaning on. Third, we are formalising monetary divergence as its own theme because it is the cleanest currently-live example of the same shock producing opposite policy: energy is disinflationary for the US through demand destruction and inflationary for Japan through the import bill. Invalidation for risk-on keeps its three legs from 08-16 and gains a fourth: minutes that reveal broad committee sympathy for the dissenters, which would reprice the front end against a market that has moved the first hike into 2027. Two-sided as required by our own recurring lesson: the equally live outcome is minutes that read as a divided committee whose hawks have already been overtaken by the data, which extends the risk-on tape without a hike to fear and makes the consumer, not the Fed, the only thing left that can break it. Watch the front end on Wednesday and the trade-down shape across the retail block. Outcome: open

  • 2026-08-18: The 08-15 call to watch the SHAPE rather than the level is now the only framing that survives contact with Monday. On one day, September hike odds fell to 30-31% and the 30-year hit 5.31%, the highest since June 2007. A market that has moved the first hike into 2027 and a long end at a 19-year high are not contradictory readings of the same economy, they are readings of two different questions, and only one of them belongs to the Fed. Score the 08-17 call on this as a HIT: we wrote that the Fed's hold is being delivered through the long end and financial conditions rather than the policy rate, and that an allocator holding duration is exposed to the instrument the Fed is currently leaning on. Monday priced exactly that. Second, the 08-16 invalidation list had three legs and the THIRD one fired: an oil disruption reaching the portfolio through the pump. Brent at $90 on the expiry of a dated deadline is a different fact from Brent at $88 on a headline, because the expiry removes the option value of a near-term resolution rather than repricing its odds. The honest qualifier, and the reason we are not calling risk-on invalidated: the equity move was -0.5% with the Nasdaq outperforming, which is a risk event being absorbed, not a regime break. Third, and this is the correction we owe ourselves, we have been building a clean demand-destruction story and Monday inserted a fact that does not fit it. Empire State printed 20.6 with prices paid at 58.6 on the same day the consumer story got global confirmation from Japan and China. Our own recurring lesson is that we name the right variable and attach a one-sided direction, so state this two-sided: the minutes on Wednesday can either reveal a committee that has been overtaken by the consumer data, which extends the tape, OR reveal hawks whose supply-shock-plus-AI-capex argument now has a regional survey and a $90 Brent print supporting it. Fourth, the new thing to watch that we have not been watching: Canada gave us the cleanest natural experiment available on whether this energy shock reaches core. Headline 3.0%, gasoline +25.7%, ex-gasoline 2.2% for a third straight month, core at 1.9%-2.0%. In July, in a G10 economy, it did not reach core. That is one country and one month, and US core PCE on 26 August is the version that matters, but it is evidence and we should update on it rather than on intuition. Invalidation for risk-on keeps its four legs and the emphasis shifts: the live one is now the long end, not the policy rate. Watch the 30-year and the front end SEPARATELY on Wednesday, and read the retail block for whether $90 Brent shows up as a cost line or a demand line. Outcome: open

Hit-rate

  • 1 hit / 1 partial miss (2 closed): 08-12 "watch the 10Y over the CPI headline" validated 08-13. 08-14 "watch retail sales and UMich, the risk is a HOT consumer read" scored a partial miss 08-15: right variable, wrong sign. The 08-13 call (oil = the remaining path back to yield pressure) partly resolved 08-14: the tail deflated on demand forecasts rather than materialising, so the risk framing was right and the risk simply did not fire. Recurring lesson to watch for: we keep naming the right variable and then attaching a one-sided direction to it. 08-16 correction applied: the consumer call for this week is written explicitly two-sided, with a stated shape test across the trade-down curve rather than a directional guess.
Linkedai-capexai-credityieldsgoldcritical-mineralsus-consumerenergy-hormuzrate-divergence

Recent research on Market Regime

No edition has singled out Market Regime yet. See the latest cross-asset read on Research.

Live figures are delayed and for analytics/education only — not investment advice, not a signal service, no buy or sell recommendation. Journaled calls include their outcomes, wins and misses. Every decision and risk is your own. DYOR.