BRAM Proprietary AI Market Selection Tool: Our New Tools and Framework to Screen Workforce Multifamily Housing Markets
Overview
Here at Ballast Rock Asset Management, our real estate team has built a proprietary tool utilizing our deep expertise and implemented using AI, to score every U.S. multifamily housing metropolitan area based on key categories. This proprietary tool allows our origination team to better source multifamily workforce housing in optimal markets. The tool gives us four key scores per market, and we weight different key scores depending on the kind of deal we are underwriting. This helps our team discern the good from the bad markets for this given type of acquisition, serving as a screener rather than an actual underwriting or predictor tool.
Why we screen instead of forecast
We started out trying to forecast metro rent growth directly, largely from publicly available data. The data is not robust enough yet to use directly in underwriting, and we do not believe it works well to project out five years. A market's rent path over the hold period for a given acquisition is mostly local and deal-specific: one large employer, a wave of new supply, or a zoning or tax change can materially move the economics of a deal on its own. Our own model produced an R2 of just 13% to 16% in testing, explaining only a small portion of what happened in the data. We feel the accuracy ceiling for this kind of forecast is low for everyone, even for those with larger or proprietary datasets.
Given the above analysis, we stopped trying to predict and built a screen instead. It sorts markets on the handful of fundamentals that have actually held up to rigorous review, organized around the four questions we ask about every deal. It will not tell us where rents are headed, but it is transparent, easy to explain, and it matches our experience of what produces the best outcomes.
What the research tells us
The screener tool is built on the factors that research most consistently links to rent performance:
Supply is the biggest near-term driver. The 2021 to 2024 apartment building wave pushed deliveries to their highest level in nearly 40 years and stalled rent growth across the oversupplied Sun Belt region. The sharp drop in construction starts since then is what sets up the next leg (CBRE).
Affordability sets the ceiling. Renter cost burden is at record high levels, with roughly half of U.S. renters now cost-burdened (Harvard Joint Center for Housing Studies). That limits how far rents can push in stretched markets and rewards the markets that still have room.
Vacancy and absorption show pricing power right now. Tight, well-absorbed markets hold rents. Markets working through heavy deliveries see vacancy rise and rents flatten, whatever the demand story.
Jobs and migration drive demand, with a catch. Metros with stronger employment and in-migration, much of it people moving to the Sun Belt for affordability, support above-average rent growth. But that same growth attracts the new supply that competes it away.
The lesson is simple: rents hold up where housing stays affordable and new supply is limited, not where growth is fastest. That is why those two factors carry the most weight in our screening tool.
The four metrics, and how we use them
Rent Persistence. How well in-place rents should hold over a five- to seven-year hold. It rewards lower in-place rents, affordability headroom and limited new supply. When the business plan is to maintain steady occupancy and high current cash flow, we look hard at rent persistence going in. It is built from rent burden measured on renter income, the in-place rent level, the forward construction pipeline as a share of stock, vacancy, and absorption. We keep demand out of our rent persistence calculation on purpose, so it reads downside durability rather than growth.
Rent Upside. How much rent growth a market could deliver if new supply gets absorbed. On a turnaround or renovation-driven deal, we put more weight on rent upside going in. We keep the two separate on purpose, because the same growth that drives upside is what attracts the supply that erodes rent persistence. It is built from employment growth, income growth, population growth, and recent class B/C rent momentum.
AI Resilience. How well positioned a metro's economy is for the AI era. We rebuilt this on hard data: each metro's job mix from the Bureau of Labor Statistics is scored against the Felten, Raj and Seamans AI exposure dataset, rewarding the concentration of computer, mathematical, science and engineering work together with broad economic diversity, on a national 0 to 100 scale across all 394 metros. The metros that score highest are diverse, knowledge-anchored economies, and the research, technology and defense hubs that lead them, such as Huntsville. Because that tilts toward larger, higher-cost metros, we treat it as a secondary lens for our workforce focus, and we read it alongside exit health to gauge buyer depth and how cleanly the asset will sell.
Exit Health. How liquid and “exit-able” a market is. On any deal, we use it together with AI resilience to set the exit cap and to judge how hard the asset will be to sell. It is built from transaction turnover, five-year sales volume, population and population growth, median income and income growth, and the prevailing cap rate.
What we are seeing
The clearest pattern is that the markets with the most durable cash flow are usually not the markets with the easiest exits. Many of the highest rent persistence metros are cheap and supply-protected, but thin and slow to trade. That is why rent persistence on its own is not enough. Places like Albany, GA, or Anniston, AL, screen well for rent durability, but we would generally pass on new opportunities today, because there is little liquidity and a narrow buyer pool at exit.
The markets we like for steady cash flow are the ones that hold rents and can still be sold: Warner Robins, Dalton, Johnson City, and Pinehurst. On the value-add and turnaround side, the stronger setups are higher-momentum, liquid markets such as Charleston, Greenville, Spartanburg, Clarksville, and Auburn-Opelika, where the tradeoff we underwrite is the supply risk to rent persistence.
A few markets balanced all four. Huntsville stands out, anchored by defense, research, and healthcare, which also makes it one of the most AI-resilient markets we cover; Gainesville, Georgia is another. More broadly, the most AI-resilient metros are anchored by universities, hospitals, defense, and research (Raleigh, Huntsville, Chattanooga), while the most exposed lean on lower-wage service and back-office work (Memphis, Albany, Ocala). We use that to temper exit assumptions where the renter base looks exposed.
Where we land versus CoStar
We do not line up with CoStar (the primary real estate industry data aggregation tool), and that is by design. Their forecast favors the high-growth, higher-cost names: The Villages, Miami, Knoxville, Palm Beach. Our screen rewards affordability, limited new supply, and a real exit. Where we see value that CoStar's growth forecast tends to underrate, it is in solid secondary markets that score well across our four metrics, such as Warner Robins, Clarksville, and Johnson City. Where CoStar is most bullish, we are often cautious, because those markets tend to fail our affordability or supply test even when the growth headline looks good. We are not trying to beat CoStar's forecast. We are screening for downside-protected cash flow, and for the places where genuine upside is worth the supply risk.
Putting it together
The hard part of market selection is that the thing we most want to know, where rents will be in five years, is the thing no one can accurately forecast with the data we have at this time. So we pivoted and built a tool for the questions we can answer: how durable is the cash flow, how much upside is realistic, how exposed is the job base, and how cleanly can we exit. We weigh those differently by deal. A cash-flow hold leans on rent persistence, a value-add leans on rent upside, and the exit is always underwritten through AI resilience and exit health together.
A clear pattern runs through the results: the safest cash-flow markets are often the hardest to exit, the most exciting growth markets carry the most supply risk, and the best opportunities are the few that balance durability with a real exit. None of this replaces deal-level work; the scores are a screen, CoStar's long-range numbers are scenarios, and every asset still demands its own underwriting. What the screen does is point us at the right markets and keeps our sourcing disciplined, moving the question from which markets are growing to which markets we can buy, hold, and sell on our own terms.
* Ballast Rock Asset Management (“BRAM”), Ballast Rock Private Wealth (“BRPW”), and Ballast Rock Capital (“BRC”) are operating entities of Ballast Rock Holdings (“BRH"), an integrated investment management company. Ballast Rock Asset Management is a non-registered entity. Ballast Rock Real Estate is a wholly owned subsidiary of BRAM. BRPW is a registered investment advisor. BRC is a registered Broker dealer and a MEMBER of FINRA / SIPC. BRC’s registered head office is 460 King Street, Suite 200, Charleston, SC, 29403. Tel: 800-204-2513. To check background information about BRC and its representatives, visit FINRA’s BrokerCheck. Please see important disclosure information in our Form CRS
