Bearish or neutral.
A bearish reading opposes the planned addition; a neutral reading gives no extra directional support. You may keep 6 years while reviewing the entry against your macro plan.
CHIRAG MIRANI / FIXED INCOME RESEARCH
U.S. 10-Year Treasury Yield Outlook
A duration view one week ahead, to use after forming your macro view.
ONE-WEEK-AHEAD OUTLOOK FOR THE 10-YEAR TREASURY YIELDIssued weekly forecast
Neutral
The two models disagree. The combined model is neutral for the stated one-week period, adding no duration tilt to the macro position.
Outlook for the week ending: close
Based on the close.
Model 1 · ChecklistBearish Duration
Score -2
5 add / 7 reduce / 2 neutralModel 2 · RegressionBullish Duration
56.6% probability of lower yields
Uncalibrated model probabilityCompare the one-week view with your macro position and risk budget.
November 6, 2026 inputs: 10-year Treasury 4.86%, 2-year Treasury 4.51%, VIX 25.64. U.S. Treasury and Cboe supplied the observations missing from FRED. Treasury values are official daily reference yields. Source details.
INSIDE MODEL 1
The checklist, grouped by direction.
Readings that favor adding duration come first, followed by reducing duration and neutral readings. These explain Model 1; the combined result above is the single signal.
| Indicator and rule | Observed state | Duration vote |
|---|---|---|
| Bullish duration readings5 readings | ||
| U.S. equitiesFour-week equity move; rising prices vote to reduce duration. | Fell over four weeks | Bullish |
| Emerging-market equitiesFour-week equity move; rising prices vote to reduce duration. | Fell over four weeks | Bullish |
| Equity volatilityFour-week VIX change; a rise votes to add duration. | Rose over four weeks | Bullish |
| Labour marketUnemployment above its preceding five-year average votes to add duration. | Above preceding five-year average | Bullish |
| Yield minus trailing inflationA nonnegative nominal yield minus trailing core PCE gap votes to add. | Yield at or above trailing inflation | Bullish |
| Bearish duration readings7 readings | ||
| OilDelayed four-week WTI move; rising prices vote to reduce duration. | Rose over four weeks | Bearish |
| U.S. dollarDelayed six-currency basket; a rise votes to add duration. | Fell over four weeks | Bearish |
| Core inflationCore PCE inflation below 2% votes to add; otherwise reduce. | At or above 2% | Bearish |
| 2s10s yield curveFour-week steepening votes to add; flattening votes to reduce. | Flattened over four weeks | Bearish |
| German yieldsA rise in the latest published monthly change votes to reduce. | Rose in latest published monthly change | Bearish |
| Japanese yieldsA rise in the latest published monthly change votes to reduce. | Rose in latest published monthly change | Bearish |
| U.K. yieldsA rise in the latest published monthly change votes to reduce. | Rose in latest published monthly change | Bearish |
| Neutral readings2 readings | ||
| December seasonalityOne reduce-duration vote during December; otherwise neutral. | Outside December | Neutral |
| 30-year TIPS auctionOne add-duration vote for an eligible announced auction in the next seven days. | No eligible auction next week | Neutral |
These are directional hypotheses. Yield minus trailing inflation is a historical gap, not a measure of fair value. The auction input uses announcements available before the reference Friday. Full definitions in the paper.
MACRO VIEW FIRST, TIMING SECOND
Bullish on bonds. How much to add now?
Suppose your portfolio has 6 years of duration and your macro view supports a target of 7. Set the target and risk budget first; use the weekly evidence to review the entry.
Bullish duration.
Consider moving toward the planned 7-year target if the portfolio conditions still fit. The checklist does not set the amount.
The same bullish reading.
If you prefer a staged entry, add 0.25 years to reach 6.25, then reassess before the next step.
Illustrative decisions, not a tested allocation rule. The model itself uses fixed +1, 0 or -1 positions.
HISTORICAL RESEARCH · NOV 2016–OCT 2026
The results, with the comparison in view.
52.9%Hit rate · 342 active weeks
+32.6Yield bp per year · before charge
+27.2After illustrative charge · bp/year
Quarterly refits and the 30-year TIPS addition were selected after inspecting the later history. The same period's always-higher-yields benchmark scores +31.4 bp/year before the charge. The selected model's four-lag t-statistic is 1.67; these are descriptive results after selection.
Scores are signed changes in the 10-year yield, not investment returns. The charge is 0.25 yield bp per position unit changed, including entry and exit. Every historical week (CSV) · Full statistics (JSON)
PROSPECTIVE WEEKLY RECORD
Track each call as it happens.
1 forecast issued · 0 outcomes recorded
Issued forecasts keep their actual timestamps. Outcomes are appended when available; historical reconstructions are kept out of this record.
0 reference weeks without a forecast recorded, including the current week if still awaiting its issue. Weekly coverage (CSV).
—Cumulative gross yield score · bp
—Hit rate · completed active calls
0Completed active calls
| Reference Friday | Duration view | Outcome | Yield score, bp |
|---|---|---|---|
| 2026-11-06 | Neutral | Pending | — |
+1 is Bullish Duration; -1 is Bearish Duration; 0 is Neutral. Unavailable weeks are not neutral forecasts. Weekly record (CSV) · JSON · Latest reading (JSON)
QUICK ANSWERS
Using the Treasury checklist
What does the Treasury signal help me decide?
It is a duration view one week ahead, used after forming a macro view. Bullish Duration favors lower yields and supports adding duration; Bearish Duration favors higher yields and supports reducing duration. Neutral adds no directional tilt. Position size remains a portfolio decision.
What does Signal generated as of mean?
It is the actual time the displayed reading was generated, shown in New York time. The one-week period is shown separately. A historical reconstruction can be generated after that period ends and is explicitly labelled; its generation time does not make it a current forecast.
Is the position scaled by the score?
No. The model outputs +1, 0 or -1. The score shows the balance of votes; the regression probability is uncalibrated. Portfolio size comes from your risk budget and implementation plan.
Can I use the page on Monday or another day?
Yes. You can check the latest published data on any day. The tested signal remains Friday to Friday; a Monday visit does not create a Monday-to-Monday forecast.
How are results measured?
The score is minus the position times the next weekly change in the 10-year Treasury yield, in basis points. Flat weeks score zero. It is a yield-direction measure, not a bond or ETF return.
Were these historical results selected after seeing the data?
Yes. Quarterly refits and the 30-year TIPS auction addition were selected after reviewing the later results. The preprint reports the full searches and benchmarks. Prospective forecasts are recorded separately.
What happens when data are late?
The page keeps the last complete reading and its original reference dates. It does not turn missing data into a neutral vote or backdate a new forecast. A new issue needs complete fresh inputs in the issue window.
Where are the paper and reproducible code?
The preprint page links the PDF, readable paper, full code package and historical result tables. The weekly tracker supplies JSON and CSV downloads.
Data refresh details
Last model check: November 7, 2026, 03:48:18 PM EST. Requested reference Friday: 2026-11-06.
The requested inputs passed the checks.
The button checks the most recently published data. It does not run a new model in your browser. The scheduled local refresh downloads inputs, validates them and publishes eligible weekly forecasts.