· Analytic
Z3937 LBR Housing Demand Composite Signal
Four-leg housing-demand composite, 40 percent Census building permits YoY, 25 percent 30-year mortgage rate change inverted, 20 percent Zillow new-listings and price-cut pulse, 15 percent UMich consumer sentiment, as a Grade B+ directional signal on CME Lumber forward returns.
At a glance
Current reading
-0.695
Housing demand weakness moderate
Signal state
Moderate NEG
phase -1, 43 days running
Sample
Daily-aligned, 25-year span
2000-01+ (CENSUS onset); full-power 2017+
Daily IC
+0.161 (p < 1e-50)
Spearman vs LBR fwd60d, n = 8,492
Non-overlap rho
+0.151 (p = 0.072)
Every-60d, n = 142, same sign, marginal
Phase +1 edge
+2.91 pp
Housing demand firing outperforms
Phase -1 edge
-3.88 pp
Housing demand collapse underperforms
Baseline LBR fwd60d
+2.24%
Unconditional mean (live window)
As of Oct 30, 2026. Updated nightly.
SIGNAL OVERVIEW
US framing-lumber demand is driven by housing starts, which are in turn driven by building permits, mortgage affordability, housing-market pulse, and consumer confidence. The single-feature UMich sentiment signal that previously targeted LBR missed Grade A on both daily and non-overlap IC and pinned at EXTREME_NEG for 459 trading days, consumer sentiment had sat at persistently low z for over two years, saturating the tanh and masking any inflection signal. This reformulated signal reads four orthogonal housing-demand channels: Census building permits (164K rows, 2000 onward, permits lead starts by one month, lead framing-lumber demand by 2-3 months), 30-year mortgage rate change inverted (affordability gate, 56-year weekly history), Zillow new-listings + price-cut pulse (housing-market momentum, 2017 onward), and UMich sentiment retained as a 15 percent overlay. The honest grade: Grade B+. Daily Spearman IC against LBR fwd60d is plus 0.161 at p-value 3.9e-50 across 8,492 trading days, very strongly significant daily, 14x stronger than the single-feature predecessor. Non-overlapping (every-60-day) IC is plus 0.151 at p-value 0.072 across 142 independent samples, same sign and good magnitude but just above the strict 0.05 bar. Phase stats are cleanly monotone with wide spread: Phase +1 (Housing Demand Firing) delivers plus 5.15 percent fwd60d, plus 2.91 pp above the plus 2.24 percent baseline. Phase -1 (Housing Demand Collapse) delivers minus 1.64 percent, minus 3.88 pp below baseline. 6.79 pp directional spread, 13x wider than the predecessor. Current reading: Moderate Housing Demand Weakness at composite minus 0.695, 43 trading days running. Housing-demand composite has run negative for 2+ months. Modest bearish-LBR lean.
LIVE CHART
LBR Housing Demand Composite Signal
Composite reads minus 0.695 as of October 30, 2026, in the Housing Demand Weakness band. Building permits YoY, mortgage rates, and Zillow pulse have all weakened. CME Lumber (orange) has traded in the \$510-525 range.
COMPOSITE
-0.695 / tanh
STATE
Housing Demand Weakness
PHASE
-1
DAILY IC
+0.161 (p < 1e-50)
INDEPENDENCE
p = 0.072 (marginal)
TODAY’S PLAYBOOK READING
STATE
Moderate Neg
STATE AND HISTORY
Housing Demand Weakness firing at composite -0.695, 43 trading days running. Phase +1 (Housing Demand Firing) delivers +5.15 percent fwd60d / 62 percent hit / +2.91 pp above +2.24 percent baseline. Phase -1 (Housing Demand Collapse) delivers -1.64 percent / -3.88 pp below baseline. Daily IC +0.161 at p<1e-50 very strong. Non-overlap rho +0.151 at p=0.072 across 142 samples - MARGINAL.
THE FOUR-LEG COMPOSITE
40 pct Census building permits YoY (direct physical demand signal, 164K rows, permits lead starts 1 mo, lead framing demand 2-3 mo). 25 pct mortgage-rate change inverted (affordability gate). 20 pct Zillow new-listings + price-cut pulse (housing-market momentum). 15 pct UMich sentiment (consumer-confidence overlay, down-weighted from the broken predecessor).
HOW TO POSITION
Primary structural lumber-demand signal on the library. Pair with CFTC lumber positioning, Canadian wildfire supply disruption (W106), and homebuilder ETF strength (XHB/ITB) for stacked confirmation. Current Weakness regime says bearish-LBR lean.
WHAT THE CHART SHOWS
The composite sits at minus 0.695 in the Housing Demand Weakness band. The blue signal line is below the minus 0.5 extreme threshold. Building permits YoY, mortgage rates, and Zillow pulse have all weakened. CME Lumber (orange) has traded in the \$510-525 range.
TOP INPUTS FIRING
- Primary composite -0.695 State Housing Demand Weakness, 43 days running.
- Four-leg composite firing bearish Census permits YoY, mortgage rate change, Zillow pulse, and UMich sentiment all contribute negative readings to the composite.
- Daily IC +0.161 p<1e-50 - Grade B+ (marginal non-overlap) Non-overlap rho +0.151 at p=0.072 across 142 samples, same sign but above strict 0.05 bar. Grade B+.
THE TAKEAWAY
What this means for you today
THE HOUSING-DEMAND CALL
Composite -0.695, Housing Demand Weakness moderate. Building permits YoY, mortgage rates, and Zillow pulse have all weakened. Modest bearish-LBR lean.
THE EMPIRICAL CONTEXT
Phase stats show clean monotone pattern. Phase +1 (Housing Demand Firing) +2.91 pp above baseline. Phase -1 (Housing Demand Collapse) -3.88 pp below baseline. 6.79 pp directional spread.
THE HONEST CAVEAT
Daily IC very strong at p<1e-50. Non-overlap rho marginal at p=0.072 (same sign, just above the 0.05 bar). Grade B+, use as structural directional read, confirm with positioning and supply-side signals. Reformulated 2026-11-01 from single-feature the University of Michigan consumer-sentiment index per Priya+Marcus 5-engine review.
THE REGIME MAP
The three states, forward returns and sample share
Computed from 8,492 trading days. Baseline LBR fwd60d: plus 2.24 percent. Grade B+ with daily IC +0.161 p<1e-50; non-overlap rho +0.151 p=0.072 (marginal, same sign). 6.79 pp directional spread.
| State | Phase | Share | Fwd 60d | HAC t | Edge vs baseline | Last firings |
|---|---|---|---|---|---|---|
| Housing Demand Firing | +1 | 13.0% (1113 d) | +1.76% | -0.48pp | Feb 2026, Dec 2025, Apr 2024 | |
| Normal | 0 | 69.8% (5969 d) | +0.28% | -1.96pp | Jul 2026, Apr 2026, Mar 2026 | |
| Housing Demand Weakness TODAY | -1 | 17.2% (1470 d) | -2.68% | -4.91pp | Sep 2026, May 2026, Apr 2026 |
The historical pattern
Signature episodes in the 25-year sample
THE TRACK RECORD
State changes
| State | Start | End | Duration | Fwd 60d LBR |
|---|---|---|---|---|
| Normal | 10/30/2024 | 01/31/2025 | 64d | +3.76% |
| Housing Demand Weakness | 02/03/2025 | 02/10/2025 | 6d | -7.23% |
| Normal | 02/11/2025 | 07/31/2025 | 118d | -6.65% |
| Housing Demand Weakness | 08/01/2025 | 08/29/2025 | 21d | +10.22% |
| Normal | 09/02/2025 | 12/02/2025 | 65d | +5.43% |
| Housing Demand Firing | 12/03/2025 | 12/04/2025 | 2d | -0.52% |
| Normal | 12/05/2025 | 01/30/2026 | 38d | -0.64% |
| Housing Demand Firing | 02/02/2026 | 02/02/2026 | 1d | +2.33% |
| Normal | 02/03/2026 | 03/18/2026 | 31d | +4.22% |
| Housing Demand Weakness | 03/19/2026 | 03/30/2026 | 8d | +14.40% |
| Normal | 03/31/2026 | 04/07/2026 | 5d | +12.09% |
| Housing Demand Weakness | 04/08/2026 | 04/08/2026 | 1d | +11.13% |
| Normal | 04/09/2026 | 04/30/2026 | 16d | +9.97% |
| Housing Demand Weakness | 05/01/2026 | 06/30/2026 | 41d | +1.22% |
| Normal | 07/01/2026 | 08/31/2026 | 43d | +3.43% |
| Housing Demand Weakness | 09/01/2026 | 10/30/2026 | 43d |
Scroll inside the table for the full history. Forward 60-day LBR is measured from each run’s start date; a reader cannot trade off a start-date they only identify in hindsight, so this is a transparency log, not an alpha claim. Asterisked (*) returns are partial. The run is less than 60 trading days old. The full 60-day comparable fills in as more data arrives.
What this measures
The signal pulls four housing-demand channels: (1) CENSUS_BPS total-US monthly building permits YoY z-score; (2) 30-year mortgage rate 12-week change inverted z-score (falling rate = positive signal); (3) Zillow new-listings YoY z and price-cut share inverted z composite; (4) UMich consumer sentiment level z-score. Weights: 40 / 25 / 20 / 15. State thresholds plus or minus 0.2 (moderate) and plus or minus 0.5 (extreme). Monthly release cadence for Census + Zillow + UMich, weekly for mortgage.
Why it matters
Why four-leg composite beats single-feature sentiment
The predecessor signal was a single-feature UMich consumer-sentiment z-score. the University of Michigan consumer-sentiment index often CONFIRMS housing cycles rather than leading them (post-Lehman and 2020 the index was coincident, not leading). It had pinned at EXTREME_NEG for 459 trading days, saturating the tanh output. The reformulated composite reads structural housing-demand through four orthogonal channels, building permits (leading physical signal), mortgage rates (affordability gate), Zillow pulse (market momentum), and sentiment (confidence overlay). Daily IC moved from near-zero to +0.161, non-overlap from p=0.72 to p=0.072, directional spread from 0.52 pp to 6.79 pp.
Why Grade B+ with marginal non-overlap
Daily Spearman IC plus 0.161 at p-value 3.9e-50 across 8,492 observations, very strongly significant daily. Non-overlap rho plus 0.151 at p-value 0.072 across 142 independent samples, same sign and good magnitude but just above the strict 0.05 bar. Grade B+ reflects daily very strong + non-overlap marginal same-sign. 6.79 pp directional spread, one of the widest in the softs library.
How to use this signal today
Primary structural lumber-demand read. Pair with CFTC lumber positioning (reversal-setup overlay), Canadian wildfire supply disruption (W106 supply-side gate), and XHB/ITB homebuilder ETFs (market-based forward housing-demand). Current Weakness regime means pay close attention to any inflection in permits or mortgage rates for upside catalyst.
GO DEEPER
Want the full depth?
The research brief walks through the four-leg composite construction, documents the 2008 GFC collapse and 2021 post-COVID boom anchor cases, explains the the University of Michigan consumer-sentiment index->composite reformulation, and discusses the marginal non-overlap significance.
Related signals
Not investment advice · The analysis on this page is informational and educational. Signal readings, historical case figures, and empirical statistics are derived from publicly available data and are provided without warranty. Hypothetical performance results have inherent limitations, including the benefit of hindsight. Readers who intend to trade should consult a licensed investment professional and see the research brief for the full disclaimer.
Last refreshed Nov 01, 2026 08:57 PM ET