Week 37 — Sep 7 – Sep 11, 2026

A quiet open, then the inflation week arrived early — and ended with the print the week was waiting for. Markets were closed Monday for Labor Day and Tuesday brought no tier-one data, so Week 3 began almost where Week 2 left off. By Thursday the tape had turned toward the inflation quadrant: WTI is $99.5 (from $91.48 last Friday, +9%) as the Strait-of-Hormuz premium persists; the 10-year is 4.93% (+15bp on the week), the 30-year at its 2007 high; MOVE has lifted to 82 from 77, the first sign of rate-vol stirring; the yen is quiet near 154. Friday's CPI then landed close enough to consensus to release the pressure: SPY 764.66, +0.8% on the day and −0.7% on the week (770.19 last Friday); VIX back to 16.2. This is rates repricing level on inflation and fiscal supply, not a funding event — SOFR sits at the rate on reserves and credit is tight.

Thursday's PPI was the first of the two prints that decide the Fed, and it ran hot. Producer prices rose 0.4% on the month and 5.4% year-on-year, from 4.8% — the fastest pace of the cycle, a tenth above consensus — with core PPI at 4.6% y/y (from 4.3%). The one soft spot: core PPI on the month was 0.2% against 0.3% expected, and the ex-food/energy/trade measure eased to 0.3%. Read together: the pipeline pressure is energy-led rather than broad, so far. Labor is unchanged in state (claims 206K, continuing 1,774K); housing is soft (existing sales 3.98M, −2% m/m) under a 6.7% mortgage rate.

Friday's CPI decided less than feared — and shifted the risk from the Fed to the consumer. Headline rose 0.4% on the month, in line, and 3.4% year-on-year, unchanged; core came in hot on the month at 0.3% (0.2% expected) but its annual rate eased to 2.4% from 2.5%. Read together: the energy pass-through is showing up in the headline exactly as the PPI said it would, core is firming at the margin, and the hawkish tail — a break higher in the annual rates — did not land. The market took it as relief. The uncomfortable half of the morning was Michigan: consumer sentiment 47.8, the low of the cycle (51 expected), expectations 45.8, and one-year inflation expectations jumping to 4.6% from 4.0%, five-year to 3.4%. Real hourly earnings are −0.3% year-on-year. Households are absorbing $100 oil and 6.7% mortgages, and they now expect the prices to stick — the pairing that turns a Goldilocks read toward Slowdown from the demand side rather than Stagflation from the supply side. A Fed that hikes into that (the one-year breakeven sits near 2.4%) is the policy-error path we flagged; a Fed that pauses on 2.4% core is the market's Friday bet. The decision is Wednesday.

The regime read holds, and the weekly recompute is leaning the way the tape is. On the monthly states we remain Goldilocks tilting toward the inflation quadrant: firm growth (+0.49), a labor market that holds (payrolls +162K, claims 206K), inflation re-accelerating on services and energy, and a hawkish Warsh Fed. The weekly gauge has moved with the week: the inflation composite is up to +0.08 as oil feeds through, and positioning eased to −0.09 (a rising VIX means a less crowded market — a contrarian tailwind, not a comfort), which lifted conviction to +0.08. Liquidity is the slow drag to watch — Fed net-liq 13-week momentum −1.7%, the global tide (Fed + ECB + BoJ) −2.4%, both ebbing and agreeing. No guardrail is tripped, but the margins narrowed: the S&P is +5.6% above its 10-month trend (from +7% last week) and credit is softening at the edge. The cut's third leg is now the funding leg — SOFR against the rate on reserves, at −0.9bp versus a +5bp trip — which reads no plumbing stress behind the yield move; Thursday's H.4.1 feeds the liquidity state that sizes conviction, not a trigger.

Methodology update — what changed under the hood, and why. Two new inputs went live this edition, each backtested before wiring in:

• MOVE joins the positioning cycle. MOVE is the bond market's VIX — the implied volatility of Treasuries. We added it because rate-market stress leads equity weakness in a way equity-vol does not: over 2007–26, MOVE's z-score leads the next quarter's S&P at −0.20, and −0.22 even after accounting for VIX (which is roughly zero at that horizon), so it carries signal we didn't already have. It enters as a graded headwind that nudges conviction, not an on/off veto — a blunt "cut when MOVE spikes" rule tested too fragile.
• MTUM enters the Aggressive equity sleeve when conviction is positive. Momentum is the one style factor that adds, but it crashes in risk-off regimes — so we gate it: when conviction is positive (it is now, +0.06), the Aggressive US-equity sleeve becomes ⅓ SPY / ⅓ MTUM / ⅓ MAGS; otherwise it stays the default 70/30. Gating captured momentum's full edge (backtest Sharpe 0.96→1.00, 2013–26, no added drawdown) while holding it only ~46% of the time — out of exactly the regimes where it hurts. Aggressive only; Balanced and Conservative are unchanged.

• Gold is trimmed — a sizing correction, not a change of view. Stacking the regime split, an unconditional +5pp "weak-dollar" tilt and the bitcoin-to-gold shift had gold at ~20% of the Aggressive book and 16% of Balanced, against a 10% historical median. A sensitivity backtest (2007–26, both profiles) showed every step down in gold improved Sharpe, return and drawdown — even with gold's exceptional 2019–26 run in the sample — and gold fell 11% in the 2022 hiking cycle, the regime we're leaning toward. The book's drawdown protection comes from the fast-cut, not from gold. So gold's share of the risk budget is scaled to 75% of the old value and the +5pp tilt now fires only when liquidity is not risk-off (the rule EM already followed). Result: gold ~12% Aggressive / ~9% Balanced, back at the median. The debasement thesis stands; the position was simply oversized. (Aggressive Sharpe 0.94→0.96, Balanced 0.87→0.88, max drawdown better in both.)

• Global liquidity now sits inside conviction — and a correction to what I said about it. Two editions ago I reported that global liquidity momentum led the S&P by about a quarter with a +0.64 correlation. That was measured on twenty months of data; on the full 2007–26 record the correlation is −0.16 and it changes sign between eras. As a predictor of returns, neither Fed nor global liquidity is reliable, and I've corrected that claim everywhere it appeared. What does hold up is the liquidity regime state — is the tide rising or ebbing — as the input that sizes risk. Swapping the Fed state for the global one in the conviction score (same weight, same band, nothing tuned) lifts the Aggressive backtest Sharpe from 0.99 to 1.10 with identical drawdown. We adopt the more conservative version: the score leans on liquidity only when the US and global tides agree (about two-thirds of the time) and stands neutral when they disagree. This week they agree — both ebbing — so nothing changes today; the rule matters the first week they part.
• The fast-cut's liquidity leg is retired and replaced by a funding leg. The old leg cut risk when Fed net liquidity's 13-week momentum fell below −2%. A threshold sweep on the 2007–26 backtest showed it did no protective work — every crash was caught by the trend and credit legs — while it fired alone in 17 months, nearly all post-crisis recoveries when net liquidity falls for benign reasons (facility roll-off, TGA rebuild); removing it slightly improved the result. Its replacement is the repo market's own stress gauge, the one Michael Howell names as his crisis barometer: SOFR minus the rate the Fed pays on reserves, 21-session average, arming at +5bp — the sign that reserves are scarce, haircuts are rising and the collateral multiplier is shrinking. Backtest: Aggressive Sharpe 1.06 → 1.07, drawdown unchanged, fires in the 2019 reserve-scarcity months and nowhere else alone. Today it reads −0.9bp: clear. (Housekeeping: the rate on reserves moves at FOMC meetings, so it is now refreshed weekly and self-checked for staleness.)
• Managed futures halved — the book's passenger, not its protector. An attribution of the 2007–26 backtest showed managed futures carrying a 21% average weight and contributing +0.3% a year; the crash protection everyone credits it with actually comes from the fast-cut. Moving half of the defensive budget's managed-futures share to T-bills lifts the Aggressive Sharpe 0.95→0.99 and Balanced 0.85→0.92, with max drawdown unchanged or better. We keep the other half as the one hedge for a crash the cut is late to (March 2020) and will revisit when the live vehicle (DBMF) has a five-year record. Live effect: managed futures ~15%→~8% of Balanced, cash up by the same.

Honesty note on this week's conviction uptick (+0.045 → +0.07): it came entirely from the positioning cycle (−0.49 → −0.22), and about two-thirds of that is the new MOVE input itself — Treasury vol at 76 reads as subdued rate stress, so adding it lifted the read by ~+0.18 — with only ~+0.09 from actual tape (VIX down to 15.7, the S&P's extension easing). Liquidity, monetary and credit are unchanged on the week; growth ticked to +0.49 as the August payrolls and wage prints (and July revisions) landed, inflation to +0.08 as oil fed through, and positioning to −0.09 as volatility rose — small moves, and in the direction the tape already said. The gauge itself is now strictly point-in-time: every monthly print enters only once it has actually been released, which is how the model is backtested too. Read the week as a recalibration, not a regime shift.

Alongside these, Global Liquidity (Fed + ECB + BoJ, Howell-style) now sits beside Fed net-liq as the leading tide. A broader style-factor rotation (value/small-cap/high-dividend) was tested and rejected — it underperformed SPY+MAGS in every era.

Positioning — the regime read says HOLD; the book moves once, for a model reason. The only trade this week is the gold trim above — proceeds go pro-rata to the other risk sleeves, so the risk budget itself is unchanged. Nothing about the tape or the CPI setup argues for a change, and none is made. Duration stays out (no long bonds into a hot-CPI/hawkish-Fed setup); the gold position is now right-sized rather than removed, and EM keeps its liquidity-gated tilt. We move from HOLD to reducing risk only if a hot CPI or a fresh oil leg-up lifts yields and either breaks the S&P's 10-month trend or leaks into credit, in the same window. Short of that, we hold and let the guardrails work — and this note updates once more on Sunday with the week's close.