Actionable insights — Fed Walking Into a "Mistake"

11:20 1. Size the surprise in basis points, then attribute it by category

1. Write the consensus and the actual to two decimals, not the rounded headline: here core 0.24 expected vs 0.29 actual — a 5 bp miss.
2. Note the rounding trap: 0.24 and 0.29 display as "0.2" and "0.3" on the terminal, which reads as a full tenth hotter.
3. Find which categories account for the miss (weight × deviation from their trend). If one or two account for all of it, the miss is idiosyncratic.
4. Classify those categories: one-off, volatile or policy-driven (a single firm's price-plan change, a platform policy shift) versus broad and persistent (shelter, services ex-housing).
5. Then list the offsets moving the other way — rents, food mean-reversion, drug prices — to judge the underlying trend rather than the print.

• Consensus to two decimals before each CPI; category contributions in the release detail; whether the flagged categories reverse the following month; moves in fed-funds futures that are out of proportion to the basis-point miss.

13:57 2. Check a strong payroll print against the unadjusted number, same month last year

1. Pull the non-seasonally-adjusted (NSA) payroll change for the month — "not contaminated by any models or any adjustment."
2. Compare it with the NSA change for the same month a year earlier.
3. Compare the two years' seasonally adjusted (SA) results. If a weaker NSA gain this year produces a much stronger SA figure than a stronger NSA gain did last year, the strength is coming from the seasonal factors, not from hiring.
4. Discount the headline accordingly and look for the revision risk.

• BLS NSA establishment-survey tables each jobs Friday; subsequent revisions to the SA figure; changes to seasonal factors after the annual benchmark.

32:05 3. Count the long-end move as tightening already done

1. Measure how far the 10-year yield has risen since the last policy inflection.
2. Convert it into fed-funds equivalents with her rule of thumb: roughly a 2-for-1 ratio — a 0.5 pp rise in the 10-year ≈ 100 bp of hikes.
3. Add that to the policy rate when judging whether policy is restrictive; a hike on top may be double-tightening.
4. Check the lag from the last time long yields reached this level: the 5% of October 2023 was followed by a rapid labour slowdown into 2024 and the August 2024 flash crash.
5. Look at the rate-sensitive sectors that respond first — housing, then small businesses, then jobs.

• 10-year yield vs its March 2026 level; housing starts and existing-home sales; small-business surveys; payroll momentum two to three quarters after the long-end peak.

19:08 4. Place the economy in the cycle before judging a hike

1. Date the cycle trough (Bloomberg Economics called it last fall).
2. Read the labour-market internals: a longer average workweek lifting real weekly earnings even as real hourly earnings fall is a sign of improving momentum — early cycle.
3. Compare with the Fed's historical pattern: it "tends to hike late cycle, not in the early to middle cycle."
4. A hike in an early-to-mid-cycle, "fragile" recovery shortens the recovery's runway rather than cooling an overheating economy.

• Average weekly hours; real weekly vs hourly earnings; unemployment-rate trend; whether the FOMC statement acknowledges cycle position.

22:09 5. Measure market liquidity by an intervention's effect per dollar

1. Pick the closest historical precedent for the operation and its size: the Fed's 2011 Operation Twist, $400bn.
2. Compute the size ratio to the new operation: $400bn vs a $4bn buyback — about 100x.
3. Compare the one-day and two-day yield effects. If the small operation moved yields by far more than 1/100th of the precedent's effect, the market is illiquid.
4. An illiquid market with "a lot of room for price discovery" is when an intervention is most effective — which is also why policymakers like thin-liquidity months such as August.

• Treasury buyback operation sizes and the long-end yield reaction on announcement day; bid-ask and market-depth measures in 10s and 30s.

23:45 6. Grade any intervention against the FX-intervention checklist

1. Disorder: is the market disorderly (not just moving against the policymaker)?
2. Surprise: did the operation come as a surprise, or was it telegraphed and priced?
3. Abundant resources: does the size prove the policymaker can keep going — did it beat what markets "had hoped for"?
4. All in: is there a credible commitment to "open the hose," so the announcement alone moves the market "before you even do it"?
5. If size disappoints expectations, expect the move to reverse — and read it as a sizing failure, not proof the tool doesn't work.

• Size of the next buyback step vs market expectations; surprise timing (off-calendar announcements); any open-ended commitment language from Treasury.

25:36 7. Read the sign of the oil–yield correlation as a safe-haven test

1. When an oil spike reflects higher geopolitical risk and uncertainty, check which way long Treasury yields moved.
2. Historically, rising uncertainty pushed long yields down as money sought the safety of Treasuries.
3. If long yields rise with oil instead, the market is pulling back from bonds — global risk aversion toward US Treasuries, not just an inflation-expectations effect.
4. Treat a sustained positive correlation as a warning about safe-haven status.

• Rolling correlation of 10-year yields with WTI/Brent; long-end yields on risk-off days; foreign official Treasury holdings.

37:15 8. Forecast an investment boom's GDP contribution from its growth rate, not its level

1. Remember that GDP growth depends on the change in investment: a category contributes to growth only while its spending keeps growing faster.
2. Estimate the current contribution (AI capex ≈ 1 pp of GDP growth in H1 2026).
3. Project next year's capex growth rate. If it slows — even with spending still rising — the contribution shrinks (to ~0.5 pp or lower).
4. Treat the peak in the growth rate (the "double derivative") as the peak in GDP support, and discount strong contemporaneous GDP prints accordingly ("right before 2008, growth was amazing too").

• Hyperscaler capex guidance growth rates (y/y), not dollar levels; the investment-in-equipment and software lines of GDP; next year's capex growth vs this year's.