8 power users showed me everything my Birmingham BRRRR screen got wrong
About two weeks ago, I posted a data-driven screen of every SFH listing under $200K in Birmingham, AL. Ran all 350 through a consistent underwriting model: bedroom-based Rentometer rents, $/sqft ARV, 20% down conventional, 75% LTV refi at 6.5%, DSCR 1.25 threshold, 8% CoC target.
Zero passed. 13 came close. 316 failed outright.
I got a lot of good feedback on that thread and wanted to follow up with what I actually learned.
1. Wrong slice of the market
Jason Cory (broker in Birmingham) said the median home price there is $210-214K. My whole dataset was the bottom half of the market. Those $80-120K properties are basically cash assets that trade between investors. Conventional financing and BRRRR refinancing don't really apply at that price point.
Crystal Smith mentioned something I hadn't even thought about: I ran the whole screen on list prices. Investor deals in this range close below list a lot. Without sold price data for these specific areas, my cost assumptions were off before I even started.
2. Vacancy was doing two jobs and doing neither well
I used a flat 8% vacancy rate across the board. Drew Sygit shared FICO default probability data. Tenants in the sub-600 range (which is who you're renting to at this price point) have a 15-40% default probability.
Masoud Arouni put it well: physical vacancy and economic vacancy are two different numbers, not one. Physical is the unit sitting empty between tenants. Economic is occupied but not paying: eviction in process, partial payments, collections. They run at the same time, not instead of each other. My model had zero economic vacancy built in.
Drew also shared his property class breakdown:
Class A, FICO 680+: 0-5% default, 5-10% vacancy, 3-5 years to positive cash flow
Class B, FICO 620-680: 5-10% default, 10-15% vacancy, 1-3 years
Class C, FICO 560-620: 10-20% default, 10-20% vacancy, immediate cash flow but low appreciation
Class D, FICO under 560: 20-30% default, 15-25% vacancy, Section 8 territory
My targets are mostly Class C sliding into D. 8% vacancy is a Class A number.
3. Rehab ratio breaks down at sub-$200K
Marcus Auerbach's point: rehab doesn't get cheaper just because the house was cheap. $25K rehab is 28% of a $90K property, 10% of a $250K one. Materials cost what they cost.
Travis Timmons said sub-$100K OOS is about the riskiest strategy a first-timer can run. Sweet spot is $200-300K where a $50-70K rehab actually gets absorbed. He also mentioned true cash needed for a $200K property runs $60-70K once you add closing costs, make-ready, and reserves on top of the down payment. More than most people budget for.
4. Might work as buy-and-holds, not as leveraged BRRRRs
This one from Tabish Masood changed how I'm thinking about the whole dataset. My numbers showed a 0.95 average DSCR with 76% positive cash flow. So the operational side actually works on most of these. They fail specifically on the refi at 75% LTV.
Run the same listings as buy-and-holds at 70% or 65% LTV instead and the pass rate is probably a lot higher. More cash left in, but viable. Changes the question from "do these work" to "do they work as leveraged BRRRRs specifically."
He also flagged that pre-1950 housing stock in the $40-90K range (foundation, knob-and-tube, cast iron) blows past rehab estimates more often than not. Running that re-analysis next.
5. Tax side
Ashish Acharya pointed out something nobody else touched on. Rehab spend can be broken out and bonus depreciated in year one instead of sitting in the 27.5-year bucket. Doesn't flip a bad deal into a good one, but marginal deals might look better after tax than the DSCR/CoC numbers alone suggest. Also, a cash-out refi is a non-event for tax purposes, and your depreciation basis is purchase price plus rehab, not the refinanced amount.
Two things I'm actually doing differently going into the next pass: getting sold price data from a local agent, and talking to a CPA about cost seg before I do anything with these numbers.
Appreciate everyone who took the time on this. Wasn't expecting that much detail from people who don't know me.