Forecast journal — 1 October 2026

ml · data as of 2026-09-29 · review by 2027-01-04, resolves 2027-10-01

Calls

Mine (data) Yours (gut) Horizon
B: expensive money, no crisis 55% 50% 3 years
C: fiscal dominance / 1970s-lite 30% 35% 3 years
A: disinflation returns 15% 15% 3 years
10-year on 1 Oct 2027 (now 5.26%) 4.25–6.50%, central 5.4 5–7% 12 months
30-year on 1 Oct 2027 (now 5.59%) 4.90–6.75%, central 5.85 6–9% 12 months

Method. Framework prior B 50 / A 25 / C 25, updated on the data room: A −10 (no growth trigger, Fed hiking), B +5 (vote model called B May–Aug), C +5 (fiscal strain, term premium). Ranges: every 12-month change since 1962 from a 4.5–6.5% level, applied to today and tilted up for C; each catches about 3 in 4.

Three reasons

  1. Real rates and the Fed drive the sell-off (B). TIPS +111bp in 12 months, breakevens flat at 2.36%, 5y5y 2.36%, Fed hiked to 4.00%, 2s30s flattened 113bp to 70bp.
  2. Fiscal and global strain keeps C at 30%. Interest 21.1% of receipts (C band), deficit 5.4% of GDP at 4.1% unemployment, Kim-Wright term premium 0.50 to 1.02, JGB 10-year 1.66% to 3.06%, oil $96.
  3. A has no trigger; buyers are present. Unemployment falling, HY spreads 3.1%, nominal GDP +6.3%, September 30-year auction 6.2bp through with indirect bidders at 79%.

Your ranges. 10-year 5–7% catches 52–61% of past outcomes; the 5% floor is too high (a fall of 26bp+ happened in 34–45% of years). 30-year 6–9% catches 13–30%; 9% needs +3.4pp, about the 1980 record. To express C, use 5.25–7.0%.

Watch

Would change my mind. Toward C: 5y5y above 2.6% for two months; TIPS falling as the 10-year rises; 2s30s above 120bp with the Fed on hold; 30-year tails above 3bp twice. Toward A: unemployment 4.6%+, HY spreads above 4.5%, core PCE below 2.5%.

Halve a probability if: 1. 5y5y at 2.75%+ for two month-ends: halve B, mostly to C. 2. Unemployment 4.6%+ and HY spreads 4.5%+: halve B; A rises to about 35%. 3. 5y5y at 2.10% or lower, ACM term premium 0.40 or lower, and 30-year indirect bidders 70%+ for three auctions: halve C.

Full entry: 2026-10-01.md. Scored in forecasts.csv.