Upfront: I work with housing-market data rather than owning rentals in NoVA, so take this as the numbers half of your stress test, not boots-on-the-ground. The most useful thing I can hand a rookie underwriting Loudoun and Fairfax is this: the standard guardrails you'll get here (the 1% rule, a flat 5% vacancy, a blanket "1.2 DSCR minimum") were calibrated on cheaper markets, and your submarket sits roughly 3x off where those rules assume it does. So the fix isn't a better rule of thumb, it's backing each guardrail out of the actual block.
Here's the block. I pulled the 2024 census-tract data for your two counties. Median-tract gross rent-to-price (annual median rent divided by median home value, a screening number, not underwriting) is about 4.4% in Loudoun County and 4.5% in Fairfax County, which is roughly a 22-23x price-to-annual-rent multiple. The 1% rule implies about a 12% gross yield (8x), so in NoVA it's off by nearly 3x and useless as a filter — half the deals that "fail the 1% rule" here are just normal NoVA. That number is the thing your DSCR and cash-on-cash floors have to respect.
What it means mechanically: at a ~4.4% gross yield, after you load in NoVA property taxes, insurance, maintenance, capex, vacancy and management, the net operating yield on a market-rent townhome lands somewhere around 2-2.5% of price, which is below the annual debt-service cost on a conventional loan at recent rates. Translated: on a median-priced Loudoun or Fairfax SFH/townhome at market rent, you generally can't reach DSCR 1.2 or positive cash-on-cash at 20-25% down — you'd need a down payment far north of that, closer to half cash or more, to get there. That's not a reason to skip the market, it's the tell that a straight market-rent rental here is structurally an appreciation bet, and your stress test should treat it as one. If cash flow from day one is the actual goal, the same data points you one county out: Prince William tracts run about 5.5% gross (roughly 18x), and Manassas Park crosses into the 6s (only a handful of tracts, so treat it as a pointer, not gospel) — that's where the DSCR math pencils at normal leverage without a house-hack.
And don't let rent growth quietly rescue a thin deal, because historically it hasn't kept up. Over 2014-2024, tract-median home value in Loudoun rose about +61% while median rent rose about +44% (Fairfax +48% vs +39%; Prince William +72% vs +39%) — value outran rent in every one of them, which is exactly why yields compressed to today's 4-5%. So stress your rent flat or +2-3%/yr, not the appreciation rate, and re-run the deal there. The demand under it is real (Loudoun's county population grew about +6.4% from 2020 to 2025, Prince William +4.2%, Fairfax a slower +1.7%), but demand has shown up in price, not rent.
On the specific floors you asked about, recalibrated to the block instead of a national default. For vacancy and turnover, don't paste in a flat 5% — pull your tract's real numbers; ACS table B25004 gives vacancy status and B25038 (year householder moved in) is a decent turnover proxy. NoVA is a tight, high-income metro (median tract income runs $155-177K in these two counties), so structural vacancy tends to be low, but a single turn on a $2,500 rent is expensive, so stress a full 1-2 month vacancy per turnover in dollars rather than trusting a small percentage. For reserves, month-count rules understate the cushion at these prices: six months of the actual PITI is a floor, but size a separate capex sinking fund off the real components (roof, HVAC, systems), because on a $700K asset one deferred-capex event dwarfs a few months of rent. For DSCR itself, don't accept a blanket 1.2 — solve for the down payment that gets your specific tract to 1.2 at true market rent and realistic expenses, then stress it at rent -10% and rate +1 point; if the required down payment comes out absurd, that's the deal telling you it's appreciation, not cash flow. And for the wholesale/flip side, the 70%-of-ARV-minus-repairs screen still works, but comp at the tract level — the decade appreciation spread inside each of these counties is wide (Loudoun tracts ran roughly +36% to +90% over that window), so a ZIP-wide ARV will mislead you block to block.
All of it reproduces free at the tract level on data.census.gov: B25064 (gross rent), B25077 (home value), B25004 and B25038 (vacancy and turnover), B01003 plus the Census population estimates (demand). Two honest caveats: tract ACS medians are small-sample 5-year estimates and skew single-family, so use rent-to-price as a relative screen between blocks, never as the underwriting itself; and pull your exact tract, because the county median hides a lot of what your specific deal is standing on.