Nimble Development
← Back to articles
Housing

Under the Hood: How the Nimble Attainability Index Works

Under the Hood: How the Nimble Attainability Index Works

Most housing affordability indices tell you what you already know: housing is expensive. The NAR Housing Affordability Index, the Atlanta Fed's Home Ownership Affordability Monitor, the NAHB/Wells Fargo Housing Opportunity Index — they're diagnostic tools. They measure the problem. They don't tell a builder where to deploy capital.

The Nimble Attainability Index (NAI) asks a different question: Is anyone building what people in this market can actually afford? And if not, how wide is the gap between what's being built and what's needed?

That distinction matters. A market can score poorly on a traditional affordability index (high prices relative to income) while scoring well on an attainability index (strong FHA eligibility, rising inventory, cooling prices). The first tells you people are struggling. The second tells you conditions are forming for a solution.

Here's exactly how we built it.

How We Sourced the Data

The NAI draws from five public data sources, chosen for coverage, reliability, and update frequency.

Census American Community Survey (ACS). The most granular free income data available at the census tract level. We use median household income, gross rent, and housing unit counts. The limitation: ACS data lags 12-18 months. The 2023 5-year estimates, the most current available, reflect 2019-2023 averages. We accept this lag because income distributions shift slowly relative to market dynamics.

Redfin Data Center. Redfin publishes weekly and monthly housing market data covering median sale prices, days on market, inventory, sale-to-list ratios, and price drops — broken down by city, ZIP, and property type. We chose Redfin over alternatives for three reasons: public access without API keys, weekly refresh cadence, and coverage of price bands by property type (single-family, condo, townhouse). The data goes back to 2012, giving us 14 years of monthly history for backtesting.

FRED (Federal Reserve Economic Data). The Federal Reserve Bank of St. Louis publishes canonical economic data via a free API. We pull the 30-year fixed mortgage rate (MORTGAGE30US) weekly. Mortgage rates are the single largest lever on affordability — a 1% rate increase reduces purchasing power by roughly 10%.

HUD FHA Loan Limits. The Department of Housing and Urban Development publishes annual FHA loan limits by county. These caps determine the maximum home price eligible for FHA financing (3.5% down payment), which is the primary path to homeownership for first-time buyers. A market where the median price sits well below the FHA limit has structurally better access to financing.

Census Building Permits Survey (BPS). The only nationwide source of building permit data by unit type at the county level. Permits are a leading indicator — they tell you what's coming 12-18 months before it hits the market. A market with rising permit activity is adding future supply.

Each source has limitations. Census income lags. Redfin coverage varies by region (smaller markets have thinner data). FRED rates are national, not local. HUD limits are county-wide, not city-specific. We designed the scoring to degrade gracefully when data is incomplete, using fallback values calibrated to each market.

The Eleven Metrics

Each component scores 0-100, where higher means more attainable (better for buyers).

1. Income-Price Gap (25%)

The most direct measure of affordability. We calculate the maximum home price a median-income household can afford at current mortgage rates using standard FHA assumptions (30% of gross income to housing, 3.5% down payment, 30-year fixed), then compare that to the actual median sale price.

If the affordable price exceeds the median sale price, the market scores 100 — homes are within reach. If the median price is 25% above the affordable price (1.25x), it scores 0. Everything else scales linearly between. The 25% threshold reflects the practical attainability frontier — the maximum gap a buyer can realistically close by stacking FHA financing, down payment assistance, rate buydowns, and right-sized product. Beyond that, the gap is structural.

2. Supply Pressure (10%)

Months of inventory — total active listings divided by monthly sales pace. The National Association of Realtors defines a balanced market as 4-6 months of supply. Below 3 months is a seller's market (low score). Above 6 months is a buyer's market (high score).

This metric captures market dynamics that income data misses. A market with rising inventory is becoming more attainable even if prices haven't adjusted yet.

3. FHA Headroom (9%)

The gap between the county's FHA loan limit and the median sale price. A $100K+ buffer means most homes in the market are FHA-eligible, opening the door for 3.5% down payment buyers. A negative gap (median exceeds FHA limit) means FHA financing is effectively unavailable for the typical home.

4. Absorption Rate (7%)

Homes sold relative to inventory. A low absorption rate (fewer than 15% of listings selling per month) indicates a sluggish market — good for buyers who have negotiating leverage. A high rate (above 50%) indicates a hot market where homes sell before buyers can react.

5. Rent-to-Price Ratio (7%)

Annual rent divided by median sale price — effectively the gross rental yield. Markets where the ratio exceeds 8% offer strong buy-vs-rent economics. Below 3%, renting is significantly cheaper than owning, which suppresses demand and signals overpricing.

6. FHA Utilization (5%)

A proxy for what percentage of sales could qualify for FHA financing. When the median price sits at 60% or less of the FHA limit, most transactions in the market qualify. When it exceeds 120% of the limit, very few do. This metric weights lower because it's partially redundant with FHA Headroom, but captures the distribution differently.

7. Pipeline (7%)

Building permits per 1,000 existing housing units over the trailing 12 months. Higher permit rates signal future supply additions. A rate of 15+ per 1,000 units scores 100 (aggressive building). Below 2 per 1,000 scores 0 (stalled pipeline). This is a forward-looking indicator — today's permits are next year's inventory.

8. Market Health (10%)

A composite of sale-to-list ratio and the percentage of listings with price drops. When homes sell below asking (sale-to-list under 0.95) and a high percentage of listings have price reductions (30%+), the market favors buyers. When homes sell above asking with few price drops, it favors sellers.

9. Price Momentum (10%)

Year-over-year change in median sale price. Negative momentum (prices falling) scores high because declining prices improve attainability. A market with -10% YoY change scores 100. A market with +15% YoY change scores 0. This captures the direction of affordability, not just the current level.

10. Connectivity (5%)

Mean commute time from the census tract to the nearest employment hub. A 20-minute or shorter commute scores 100. A 60-minute or longer commute scores 0. Commute cost is a real housing cost — a $350K home with a 2-hour daily round trip is less attainable than it appears once you factor in fuel, vehicle wear, and time. Data comes from Census ACS commuting characteristics.

11. Equity Disparity (5%)

The homeownership rate gap between white households and the largest minority group in the tract. A gap of 5 percentage points or less scores 100. A gap of 30 percentage points or more scores 0. Structural barriers to ownership — redlining legacies, lending disparities, wealth gaps — show up in this metric. Markets where homeownership is broadly accessible across demographics score higher. Data comes from Census ACS tenure by race tables.

The Affordable Price Formula

The anchor calculation behind the Income-Price Gap metric:

Step 1: Monthly housing budget = (Annual household income × 30%) / 12

Step 2: Monthly loan payment capacity = Budget minus property tax and insurance (estimated at 2% annually)

Step 3: Maximum loan amount = Payment capacity / mortgage payment factor (30-year fixed at current rate)

Step 4: Maximum home price = Loan amount / 0.965 (reflecting 3.5% FHA down payment)

Worked example at $85,000 income, 6.8% mortgage rate:

  • Monthly budget: $2,125
  • After tax/insurance estimate: ~$1,790/mo available for P&I
  • Mortgage factor at 6.8%: 0.00652
  • Max loan: ~$274,500
  • Max price (3.5% down): ~$284,500

If the median sale price in that market is $380,000, the gap is 34% — meaning the typical home costs a third more than what the typical household can afford with FHA financing.

How We Established the Weights

We want to be transparent about this: the current weights are informed heuristics, not empirically optimized parameters.

Income-Price Gap carries the heaviest weight (25%) because it's the most direct measure of whether people can afford what's being sold. In v4, we elevated this from 20% and tightened the scoring — the 0-point moved to 1.25x affordable (25% overshoot). This reflects the practical ceiling: even with every available program stacked (FHA + DPA + buydowns), a gap beyond 25% is structural and unsolvable by the buyer alone.

Supply Pressure and Market Health each carry 10%. Inventory dynamics and transaction-level signals (sale-to-list, price drops) capture whether the market is moving toward or away from buyers. Price Momentum (10%) captures the direction of change.

FHA Headroom (9%) reflects financing accessibility. Pipeline, Absorption, and Rent-to-Price each carry 7%, reflecting their role as supporting signals. FHA Utilization (5%) is partially redundant with headroom but captures distribution differently.

Connectivity (5%) and Equity Disparity (5%) are new in v4. Commute burden is a real housing cost that traditional indices ignore — a $350K home with a 90-minute commute is less attainable than the price suggests. Equity disparity captures structural barriers: markets where homeownership is broadly accessible across demographics are genuinely more attainable.

The threshold values within each scoring function draw from established benchmarks. The "6 months = balanced market" standard comes from NAR's widely-cited definition. The 30% debt-to-income ratio is the FHA qualifying standard. The 8% gross yield threshold for rent-to-price comes from real estate investment analysis conventions.

We acknowledge the gap: these weights reflect our judgment about which factors matter most, informed by experience and industry research, but they haven't been validated against historical outcomes. Which is exactly why we built the learning system.

How the Algorithm Learns

The NAI is designed to improve over time through a process we call Recursive Self-Improvement Learning (RSL). The core idea: the system evaluates its own predictions against real outcomes, discovers better configurations, and proposes updates for human review.

The mechanism:

We have 14 years of monthly Redfin data for our target markets — roughly 880+ city-month observations. For each historical month, we can retroactively compute what the NAI would have scored using any set of weights, then compare that prediction to what actually happened in the market 3-6 months later.

"What actually happened" is measured by a Buyer Outcome Score — a composite of realized market conditions:

  • Did homes sell (sales volume)?
  • Did buyers get discounts (sale-to-list ratio below 1.0)?
  • Did they have time to decide (days on market)?
  • Were sellers cutting prices (price drops)?
  • Was there adequate supply (months of inventory)?
  • Were prices stabilizing or falling (year-over-year change)?

A market where all six conditions favor buyers scores high. A market where sellers dominate scores low.

The validation loop:

The system computes the correlation between NAI scores at time T and Buyer Outcome Scores at time T+3 months (a 3-month lead). If high NAI scores consistently precede favorable buyer outcomes, the weights are predictive. If not, they need adjustment.

A grid search across weight combinations (at 5% resolution, constrained to sum to 1.0) identifies which configuration maximizes the predictive correlation. The system produces a quarterly report comparing the current weights against the empirically optimal weights, with correlation metrics and sensitivity analysis.

The governance:

Weights never auto-update. The quarterly report is a proposal. A human reviews the data, evaluates whether the optimization reflects genuine signal or overfitting, and decides whether to activate the new weights. Old weight configurations are preserved for audit.

This is genuine self-improvement — each cycle starts from a better baseline than the last. But it's bounded (weights converge toward optimal values) and governed (human approval gates every change). The principles borrow from online learning theory, Bayesian updating, and the kind of time-series backtesting common in quantitative finance.

What the Scores Mean

| Score | Band | Interpretation | |-------|------|----------------| | 75-100 | Highly Attainable | Market conditions strongly favor buyers. Affordable homes available, ample supply, FHA-eligible. Active opportunity for attainable development. | | 55-74 | Attainable | Solid conditions for buyer entry. Some friction (inventory tightness or price pressure) but fundamentals support affordability. | | 35-54 | Moderate | Mixed signals. Some components favorable, others challenging. Selective opportunities exist but require deeper analysis. | | 15-34 | Constrained | Significant affordability barriers. Limited supply, prices well above affordable thresholds, or weak pipeline. | | 0-14 | Crisis | Market is effectively unattainable for median-income households. Major structural barriers to affordability. |

A score is a starting point, not a verdict. A city scoring 55 might have one ZIP code at 72 and another at 38. The NAI supports drill-down to ZIP and census tract level to reveal where attainability concentrates within a market.

Try It Yourself

Our public attainability calculator covers 8 cities in California's Central Valley. Enter a city to see its current NAI score, component breakdown, and how it compares to neighbors.

For builders, lenders, and housing authorities interested in NAI data for additional markets or custom analysis, get in touch.

Sources and References

Data Sources

Housing Indices Referenced

Research

← All articles