My agent said "you're not buying a house, you're buying a rent roll with a roof attached" and I keep turning it over

The assumption doing the most work in that deal was not the capex estimate. It was the implicit cap rate floor you were applying without naming it, the threshold below which you would not buy regardless of rent stability. At 310k with 2,350 gross, you were looking at roughly a 9.1 percent gross rent multiplier. That is not a bad number for Providence in the current environment, and if you were losing the deal to deferred maintenance optics rather than a failure of that math, your agent was reading you correctly.

The deferred maintenance concern is real but separable. Siding is a known cost. You can estimate it, adjust the offer, or escrow for it. What you cannot manufacture after close is a rent roll in a market that holds. If you believed the rents were real and durable, the right move was to price the capex into the offer, not abandon the GRM entirely. That is the structural error she was pointing at.

The risk you did not mention is the one I would focus on now. A two-family in Providence at that price point likely has thin enough margin that a vacancy during a capex cycle compresses cash flow hard. The question was never whether the siding was bad. The question was whether your reserves could carry a vacancy and a capital event in the same twelve months. If the answer was yes, the pass was emotional. If the answer was no, the pass was correct but for a different reason than you gave yourself.

The separation you are describing, between a primary residence evaluation and an income property evaluation, is the actual work. You said you approached it wrong. The more precise version is that you applied a qualitative veto to a quantitative decision after the numbers had already cleared your threshold.

What was the capex estimate you were carrying in your head on the siding, and had you built that into an adjusted offer or just held it as a reason to walk?

The siding was doing emotional work, not analytical work, and that is a different problem than underwriting. Exterior deferred maintenance on a two-family in a cold-weather market is a capital event you can price, and at 310k with 2,350 in rent, there was room to quantify it and either adjust the offer or walk with a number instead of a feeling.

What nobody has flagged yet: the rent-roll framing your agent used is exactly right, and the specific risk inside it for Providence is tenant turnover cost on a two-unit. When one unit turns in a small multifamily you lose half your gross rent, and if the market rent is real and the unit needs paint and cleaning, you are looking at one to two months vacant plus leasing costs on a property where your debt service does not pause. At 6.875 on 310k your monthly payment is somewhere around 2,040 principal and interest, so a single vacancy event in year one burns most of your annual cashflow. That is worth modeling before closing, not as a reason to pass, but as a reserve requirement you bring to the table.

What did the leases look like, month-to-month or term, and do you know how long the current tenants had been in place?