Economic interpretation of all 27 readings
November 7, 2026

The thirteen main votes are retained as stated hypotheses. A sign can have a
reasonable economic explanation without having a demonstrated forecasting edge.
For every rule, distinguish the information at Friday's cutoff from the surprise
that originally moved prices and from the following week's yield change.
The data do not directly observe investors' expectations or identify a shock.

MAIN THIRTEEN

1. US equities: four-week rise -> reduce duration.
   Intended reasoning: stronger equity prices can accompany easier financial
   conditions or a stronger growth outlook, which may support higher yields.
   Limit: stock gains do not demonstrate money leaving bonds. Discount-rate
   declines can lift both stocks and bonds. Nasdaq is not a complete measure
   of US financial conditions. This is a continuation hypothesis to test.

2. Emerging-market equities: four-week rise -> reduce duration.
   Intended reasoning: stronger global demand/risk appetite may spill into US
   rates. Limit: local policies, currencies and market composition also matter;
   the index is not an observed capital-flow or Treasury-demand measure.

3. WTI: four-week rise in delayed spot observations -> reduce duration.
   Intended reasoning: persistent energy inflation can raise expected inflation
   or the expected policy path. A supply shock can also weaken real activity.
   If yields already reacted or overshot, a subsequent reversal is possible.
   Neither continuation nor reversal follows mechanically from the oil sign.
   The 2026 selected reverse direction is disclosed separately from this vote.

4. Dollar basket: four-week rise in delayed observations -> add duration.
   Intended reasoning: currency appreciation can tighten financial conditions
   and moderate price pressure. Limit: a higher expected US policy path can
   lift both the dollar and Treasury yields. No Treasury buying is observed.
   The six-currency basket is a fixed research measure, not a broad trade index.

5. VIX: four-week rise -> add duration.
   Intended reasoning: higher perceived equity risk may support demand for
   Treasuries. Limit: VIX measures option-implied equity volatility, not a bond
   order flow. Inflation risk or a need for cash can produce other outcomes.
   Its standalone historical result does not establish this sign as reliable.

6. Unemployment above the mean of 60 earlier monthly readings -> add duration.
   Intended reasoning: labor-market weakness relative to recent history may
   favor a lower future policy path. Limit: a rolling mean is not the natural
   rate or a measure of the employment gap. Participation and structural changes
   are not captured. The level rule is not an employment-report surprise.

7. Core PCE below 2% -> add duration; otherwise reduce.
   Intended reasoning: a broad policy-background threshold. The Fed's formal
   2% goal concerns headline PCE, and its numerical goal was announced in 2012.
   Applying 2% before 2012 is a retrospective research convention. A core-PCE
   threshold does not forecast the next release's surprise or the next Fed move.
   A constant sign has zero standalone timing contribution, yet can change the
   aggregate checklist, while the continuous inflation input can still vary.

8. Nominal 10-year yield below trailing core inflation -> reduce duration.
   Retained interpretation: a nominal-yield comparison with recently experienced
   inflation. The author chose to keep the rule and narrow its explanation.
   It is not a TIPS real yield, an ex-ante real yield, a term premium, or proof
   of rich/cheap valuation. A ten-year yield and past one-year inflation also
   have different horizons. The direction remains an empirical hypothesis.

9. Four-week 2s10s steepening -> add duration.
   Intended reasoning: a larger relative yield farther out the curve may make
   duration more attractive; carry and roll are the motivating idea.
   Limits: a slope change is not a carry/roll return calculation. Carry requires
   instrument income/funding assumptions; roll depends on the local curve and
   a bond aging into another maturity. A fixed ten-year reference yield does
   not measure that return. Steepening can occur with yields rising or falling,
   and an inverted curve can become less inverted. It does not prove Fed easing.
   The forecast test is still whether the next ten-year yield change is lower.

10-12. German, Japanese and UK monthly yield rises -> reduce duration.
   Intended reasoning: common rate forces or spillovers may continue into US
   yields. Limits: these are archived monthly averages received with a lag,
   not same-day foreign market changes. Same-time co-movement is not a lead.
   No foreign holdings, purchases, FX hedge costs or direct flow data enter.
   A reverse selected sign may be a reversal candidate; it needs its own test.

13. December -> reduce duration; other months neutral.
   Intended reasoning: a year-end seasonal effect. Reduced liquidity or dealer
   capacity might change the magnitude of moves, but does not determine whether
   yields rise. This code measures the calendar, not liquidity or forced sales.
   The later sample has 45 active weeks from only ten separate Decembers.
   The January-October 2026 ranking cannot evaluate December at all.

FOURTEEN ADDITIONAL READINGS
Directions in feature_hunt.csv are selected using early outcomes. They are not
economic signs fixed independently of that selection. Names do not identify
the mechanism or guarantee that the selected rule follows the named trend.

14. WTI year-on-year: an energy-price backdrop. A twelve-month change does not
    measure a new inflation shock at the weekly cutoff. Continuation/reversal
    must be distinguished from the much shorter target horizon.
15. Retail gasoline: an energy/consumer-price reading with a published survey
    date. It is not a survey of inflation expectations or a clean demand shock.
16. Natural gas: weather, storage, production, trade and demand can move this
    price. The selected 2026 sign is a reversal candidate, not an inflation law.
    The supplied follow-up tables do not identify a stable overreaction channel.
17. Oil volatility (OVX): option-implied uncertainty about oil prices. A rise
    does not specify whether oil prices or Treasury yields should rise or fall.
18. Ten-year breakeven: market inflation compensation. Risk premia and relative
    TIPS liquidity also enter; it is not a pure expected-inflation observation.
    Its construction includes nominal yields, so a contemporaneous relationship
    with nominal yields is partly mechanical. The forecast horizon is separate.
19. Fed total assets: a balance-sheet state, not a direct measure of Treasury
    net supply or an identified policy shock. Purchases, loans, repayments and
    other components differ economically. Asset changes can respond to stress.
20. Yield versus its 300-day average: a trend-state reading. The early search
    selects reduce duration when yields are below the average: that is a bet
    on a yield rebound, not a claim that the lower-yield trend will continue.
21. Yield washout above a ten-day low: an unusually large recent yield rise
    scaled by recent absolute daily changes. The early selected add direction
    is a mean-reversion hypothesis. The move need not be economically excessive.
22. Yield close near the week's high: an intraweek price-pattern observation.
    It can motivate continuation or reversal tests, but contains no direct
    measure of selling pressure, dealer inventories or event surprises.
23. Fed decision next week: scheduled event exposure. It gives no information
    about the sign of the decision surprise. Any buy/sell sign is selected from
    historical associations, not inferred from the existence of a meeting.
24. First-Friday proxy: a deterministic date rule, not the exact jobs-release
    calendar and not a forecast of payroll strength. The proxy can miss shifts
    in actual release dates. Its selected sell sign is an empirical association.
25. Fed-week event plus rising VIX: an interaction of an event flag and equity
    volatility, not evidence of a hawkish/dovish surprise or known positioning.
26. Fed/jobs proxy plus rising VIX: the same limitation with a broader event
    flag. These interaction states are sparse and overlap other candidates.
27. VIX above three-month implied volatility: a volatility-term-structure
    state. It is not a Treasury yield curve or an actual scheduled event.
    The original search groups it with event rules as a heuristic constraint;
    that grouping must not be read as an economic identity.

INTERPRETING THE COMBINATION
All votes use the same information set as the regression, and correlated votes
are not independent confirmations. Regression coefficients are conditional
associations and may change sign as inputs overlap. A 50% direction threshold
does not maximize expected yield-score profit when up/down moves differ in size.
The zero-change outcome is classified as not down, rather than as a separate
up outcome. Neither score magnitude nor uncalibrated probability determines a
validated duration size. The yield target also omits portfolio curve exposures,
carry, financing, implementation timing and costs.

Removing an input and refitting changes both model components and potentially
all regression weights. Those effects are not additive causal attribution.
A constant macro vote can affect the combined decision even when its own timing
score is zero. Such sensitivity is different from explaining market causes.

Primary background sources (definitions and mechanism context, not proof that
the checklist's weekly forecasts work):
Federal Reserve: https://www.federalreserve.gov/monetarypolicy/timeline-statement-on-longer-run-goals-and-monetary-policy-strategy.htm
New York Fed: https://www.newyorkfed.org/newsevents/news/research/2008/rp080908
Federal Reserve: https://www.federalreserve.gov/econresdata/notes/feds-notes/2016/macroeconomic-sources-of-recent-interest-rate-fluctuations-20160602.html
Cboe: https://www.cboe.com/tradable-products/vix/term-structure
EIA: https://www.eia.gov/energyexplained/natural-gas/factors-affecting-natural-gas-prices.php
CFA Institute: https://www.cfainstitute.org/insights/professional-learning/refresher-readings/2026/yield-curve-strategies
San Francisco Fed: https://www.frbsf.org/research-and-insights/publications/economic-letter/2004/01/us-monetary-policy-an-introduction-part-3-how-does-monetary-policy-affect-the-us-economy/
New York Fed: https://www.newyorkfed.org/medialibrary/media/research/staff_reports/sr570.pdf
