Neighborhood Appreciation Rates and 2026 Forecast

Neighborhood appreciation rates vary far more at the census-tract level than metro-level averages suggest. Across the 7,233 scoreable tracts in the 13 curated launch metros, the Verus-AI model identifies a 10th-to-90th percentile spread in five-year forecast appreciation that spans from single-digit cumulative gains to the model's forecast cap of 37.01%. That dispersion is the central finding: selecting a metro is a necessary but insufficient condition for capturing above-average appreciation; the tract-level signal carries the weight.

Tract 4158 in Dallas-Fort Worth leads that set with observed appreciation of 1,262.79% over the decade, a figure that is the highest observed appreciation among the 7,233 scoreable tracts across 13 major U.S. metros. Yet its Verus-AI score of 55 and grade of C- illustrate a pattern that recurs throughout the data: extreme historical appreciation does not reliably translate into a high forward-looking score. The model's handling of elevated base values is visible in the forecast chart for Tract 4158, where the 80% confidence band widens dramatically through 2029.

At the metro level, Phoenix carries the highest mean Verus-AI score among the 13 curated launch metros at 64.02, and its mean five-year forecast of 35.32% is also the highest mean five-year forecast among the 13 curated launch metros. Houston sits at the opposite end of the mean-score distribution among those same 13 metros, with a mean Verus-AI score of 51.18 and a mean five-year forecast of 28.59%. The gap between those two metros is analytically meaningful and is discussed in detail in the sections that follow.

The Verus-AI score is a composite index scaled from 0 to 100, derived from a training window spanning 2014 to 2024. The forecast horizon runs from 2025 through 2029, a five-year window. Confidence bands are 80% intervals, meaning roughly one in five outcomes may fall outside the displayed range; they are not guarantees of any particular outcome.

Coverage is broad: the model scores 82,179 U.S. census tracts across 3,234 counties, representing 84.09% of all U.S. tracts. The 13 curated launch metros account for 7,233 of those scoreable tracts. A forecast cap of 37.01% applies; 23,955 tracts across the 82,179 scoreable U.S. census tracts reach that ceiling, which means the cap is binding for a material share of the scored universe and compresses the upper tail of the forecast distribution. The national forecast distribution illustrates this compression directly: the 75th and 90th percentiles both sit at 37.01%, while the median is 30.48% and the 10th percentile is 15.7%.

Score percentiles across the 82,179 scoreable U.S. census tracts run from a 10th percentile of 37.0 to a 90th percentile of 68.0, with a median of 54.0. These figures provide the baseline against which individual metro distributions should be read. A metro whose median Verus-AI score exceeds 54.0 is, by that measure, above the national median among the 82,179 scoreable U.S. census tracts; Phoenix, Nashville, Las Vegas, and Austin all clear that threshold.

The ranked table below orders tracts by observed 2014-to-2024 appreciation. The top of that list is dominated by Dallas-Fort Worth, which places five tracts in the top ten among the 7,233 scoreable tracts across 13 major U.S. metros. Tract 4158 leads at 1,262.79%, followed by East Arlington in Jacksonville at 1,071.41% and Tract 4159 in Dallas-Fort Worth at 966.82%. Las Vegas contributes Tract 6700 at 833.75%, and Phoenix contributes Tract 3615 at 818.54% and Tract 9706 at 730.64%. Buckhead in Atlanta appears at 671.85%, and three additional Dallas-Fort Worth tracts cluster between 653.76% and 654.31%.

A pattern worth noting is the disconnect between observed appreciation rank and Verus-AI score. Among the 15 fastest-appreciating tracts in the ranked table, scores of 55 appear repeatedly, for Tract 4158, East Arlington, Tract 6700, Tract 3615, Tract 9706, Buckhead, Tract 0802, Tract 2001, Tract 1402, Tract 0100, and Montopolis.

The forward-looking, score-based view looks quite different from the historical-appreciation ranking. Nashville's strongest tract by Verus-AI score, Tract 1900, scores 85 with a 37.01% forecast. The score-based view therefore surfaces a different population of tracts than the historical-appreciation ranking, and the two should be read as complementary rather than redundant.

The most analytically interesting feature of the dataset, across the 13 curated launch metros, is the divergence between observed historical appreciation and forward-looking score. Tracts that appreciated sharply during the 2014-to-2024 training window frequently carry moderate or below-average Verus-AI scores, while tracts with high forward-looking scores often show more measured historical appreciation. This divergence suggests the model is not simply extrapolating momentum; it is incorporating factors that temper the outlook for tracts where base values have risen substantially.

Phoenix's forecast distribution illustrates one end of the spectrum. Its 10th-percentile forecast of 31.58%, the highest 10th-percentile forecast among the 13 curated launch metros, means that even the lower-scoring tracts in Phoenix's 998-tract universe project cumulative gains above 31% through 2029 under the model's central scenario. The 25th through 90th percentiles all sit at the 37.01% cap, indicating that the cap is binding for the majority of Phoenix tracts. Phoenix's mean Verus-AI score of 64.02 and mean five-year forecast of 35.32% are both the highest among the 13 curated launch metros.

Houston presents the sharpest contrast among the 13 curated launch metros. Its 10th-percentile forecast of 14.0% is the lowest 10th-percentile forecast among the 13 curated launch metros, and its median forecast of 32.21% falls well below the cap. The mean Verus-AI score of 51.18 and mean five-year forecast of 28.59% are the lowest among the 13 curated launch metros. Houston's weighted median value of $272,400 is also among the lower values in the dataset, which might lead a consensus observer to expect stronger percentage-gain potential; the model's output does not confirm that expectation, suggesting that factors beyond price level are weighing on the forward score.

Higher weighted median values do not map cleanly onto higher mean forecasts across the 13 curated launch metros. Austin, with a weighted median value of $454,400, posts a mean forecast of 33.18%, below Phoenix's mean forecast of 35.32% despite Phoenix's lower weighted median of $427,690. Raleigh carries the second-highest weighted median at $446,667 among the 13 curated launch metros and a mean forecast of 34.05%, also below Phoenix. San Antonio, with the lowest weighted median at $244,000 among the 13 curated launch metros, posts a mean forecast of 31.91%, above Houston's 28.59% despite Houston's higher weighted median of $272,400.

Atlanta's distribution warrants specific attention. Its 10th-percentile forecast of 17.32% and 25th-percentile forecast of 26.43% indicate meaningful left-tail risk relative to metros like Phoenix and Nashville. Its median forecast of 36.66% is close to but below the 37.01% cap, and its mean Verus-AI score of 52.65 is the second-lowest among the 13 curated launch metros. The wide intra-metro dispersion in Atlanta suggests that tract selection carries more consequence there than in metros where the distribution is more uniformly elevated.

The structured dataset underlying this analysis does not include rental yield figures, so no yield-based rankings or yield-to-price ratios can be reported here. What the data does provide is current home values at the tract level, which a practitioner can use as a denominator when applying external rental estimates.

The relationship between current value and forward appreciation score is not monotonic in this dataset. Among the model's strongest-scoring Austin tracts, Round Rock (score 82, current value $267,100) and Taylor (score 81, current value $275,000) sit at the lower end of the value range while carrying B+ grades and 37.01% forecasts. Shady Hollow (score 81, current value $734,402) carries the same grade and forecast despite a value nearly three times higher. For practitioners who weight yield alongside appreciation, the lower-value, high-score tracts may present a different risk-return profile than the higher-value tracts with equivalent forward scores, though any yield calculation requires rental data not present in this dataset.

The model's strongest-scoring San Antonio tracts show a similar pattern. Alamo Heights (score 77, current value $157,362) and Tract 2201 (score 80, current value $164,400) carry B and B+ grades respectively, both with 37.01% forecasts, at values well below the San Antonio weighted median of $244,000. Timberwood Park (score 76, current value $412,200) sits at more than twice the value of those tracts with a lower score. The current-value column in the ranking above is therefore a useful secondary filter for practitioners who are constructing a yield-adjusted view, even in the absence of explicit rental data in this dataset.

Because 23,955 tracts across the 82,179 scoreable U.S. census tracts reach that ceiling, a forecast of 37.01% does not distinguish between a tract that barely clears the cap and one that the model would otherwise project far higher. The cap compresses the upper tail and means that the Verus-AI score, not the forecast percentage, is the primary differentiator among high-performing tracts.

The forecast chart for Tract 4158 illustrates the extreme case: the terminal upper band for 2029 is $13,510,794, against a terminal lower band of $109,260 and a central forecast value of $1,214,985. The terminal band width of $13,401,534 reflects the model's uncertainty around a tract that appreciated 1,262.79% in a decade; the historical volatility embedded in that series produces wide forward bands. Tracts with more stable appreciation histories show correspondingly narrower forward bands.

Tracts that appreciated sharply during the 2014-to-2024 training window may carry elevated base values that compress future percentage gains, and the model's handling of that dynamic is reflected in the moderate scores assigned to many of the historically fastest-appreciating tracts. A Verus-AI score of 55 on Tract 4158, despite its 1,262.79% observed appreciation, the highest among the 7,233 scoreable tracts across 13 major U.S. metros, is a direct expression of that compression effect.

That window encompasses a period of unusually broad price appreciation in many Sun Belt markets; the model's parameters reflect that environment, and outcomes in a materially different rate or demand environment could fall outside the 80% confidence bands. Climate, insurance, and lending conditions are not explicitly modeled in the scores presented here, and no characterization of those risk factors beyond what the data labels provides is appropriate. Readers should treat the Verus-AI forecast as one input among several in any capital allocation decision.