How the Data Room is built

Fourteen monthly series span four blocks. Labor: nonfarm payrolls, average weekly hours, the quits rate, and initial jobless claims. Credit: bank lending standards from the Federal Reserve's senior loan officer survey, investment-grade and high-yield option-adjusted spreads, and real bank loan growth. Rates and the curve: the policy rate, the 2s10s slope, and the real 10-year yield. Activity and demand: industrial production, real retail sales, and new orders. Leading series (claims, spreads, the curve, orders) are aligned to lead; coincident series anchor the level.

Each series is converted to a standardized score against a trailing ten-year window and winsorized at the 1st and 99th percentiles, so a single outlier print cannot swing the index. Scores are aggregated into the four block readings and then into one composite using weights held fixed between annual reviews and published when they change. Series are aligned to their typical publication lead or lag before aggregation, so the composite reflects information as of a common point rather than mixing stale and fresh prints.

The standardized composite is then mapped back onto the familiar index scale. It is re-anchored so the reading equals 100 in January 2019, and scaled so that one standard deviation of the historical composite corresponds to a fixed number of index points. That preserves the standardized internals (every input still enters in standard deviations) while keeping a level like 119 directly interpretable against the 2019 baseline.

From the latest reading the model projects an eighteen-month path under three scenarios: a base path that carries current momentum forward, a bull path that assumes credit reopens and real income reaccelerates, and a bear path that prices a credit-led downturn. The cone widens with the forecast horizon, and in the production model it widens further as the fourteen inputs disagree: a conflicted signal reads as a wider band rather than false precision.

The composite maps to a regime band from recovery through expansion, late cycle, and contraction. Model confidence reflects how tightly the inputs agree. The twelve-month recession reading is a calibrated probability (a logistic mapping from the composite's level and slope, fit against the NBER's dated recessions), not a separate forecast. The Data Room snapshot mirrors whichever scenario is selected on the chart.

The construction is benchmarked against the established cycle composites (the Conference Board's Leading Economic Index, the Chicago Fed's National Activity Index, and the OECD's composite leading indicators) and the recession mapping is validated against NBER business-cycle peaks and troughs. Where the Composite diverges from those benchmarks, the divergence is documented rather than tuned away.

Four drivers carry the model. Rate differentials: the front-end and real-yield gap between the United States and each partner. Positioning: speculative and dealer dollar exposure measured against its own range, drawn from futures positioning and survey data. Terms of trade: commodity and trade balances that move with the global cycle. Valuation: each currency's deviation from a slow-moving fair-value anchor that pulls extreme levels back over time.

Each driver is converted to a standardized score on a common scale and combined into one reading for the broad dollar and one for every economy on the world map. Rate differentials and positioning dominate at short horizons; valuation and terms of trade carry more weight as the horizon lengthens. Weights are fixed between reviews, so any call can be decomposed back into its four drivers.

The headline figure is a conviction score from 0 to 10 that measures how strongly the four drivers align behind the current call, not how large a move is expected. A 9.7 means the drivers are nearly unanimous: rates, positioning, terms of trade, and valuation all leaning the same way, so the direction is high-confidence even when the magnitude is not. A low score flags a conflicted setup where the drivers disagree and the dollar is more likely to range than to trend.

In the Data Room, the currency-versus-dollar lens colors each economy by its year-to-date move: cool tiles are holding or gaining against the dollar, warm tiles are giving way. The equity and policy-rate lenses re-color the same map to show local-currency returns before translation and the carry each central bank is offering: alternate views of the same economies, not separate model outputs. The broad-dollar figure is the trade-weighted aggregate the individual tiles roll up to.

Three layers. Hazard combines wind, flood, wildfire, and heat exposure from federal and specialist climate data. Exposure is the insured replacement value of property sitting under that hazard. Market tracks homeowner premium growth, the coverage gap (the share of at-risk value carrying no private coverage or none at replacement cost) and the migration of risk into state insurers of last resort.

Hazard is weighted by where insured value actually sits, so concentration, not just frequency, is what re-prices a state: a moderate hazard over dense, high-value property can score above a severe hazard over empty land. The three published lenses share the same state spine but are distinct measures, not transforms of one another: the climate-risk score is the hazard-by-exposure composite, while premium growth and the coverage gap are read directly from insurance-market filings and residual-market data.

Warm tiles carry the steepest curves; the steepest-curves list ranks the states leading each metric. The national median premium sits a few points above shelter inflation and is accelerating. As private carriers retreat, the coverage-gap lens shows where the uninsured share of value is climbing and where the liability is transferring onto public balance sheets through FAIR plans and other residual markets. Read together, the three lenses show where the hazard, the price, and the retreat of private capacity line up.