Every portfolio is already selling, and the only question left is which homes go.

What does a release list built on cost leave out?

The homes on a release list are chosen on operating grounds: lease expiry, turn cost, capex plan, submarket exit, liquidity need. Every one of those is a statement about what the home costs now. Two homes with identical NOI, an identical capex plan and identical lease timing are interchangeable on that list today, and one of them will run years behind its own market while the other runs ahead.

A rating agency models zero appreciation and an AVM prices the home today, so the forward question never enters the release decision. It is answered by default, with the market rate, and the market rate is the same for every home in the market.

For a holder that is a net seller, the release list is the one portfolio lever that decides what the book looks like in a decade. A rule that cannot see forward appreciation sells some of the homes it should keep and keeps some it should sell, every quarter.

How does forward appreciation enter the release criteria?

As one more column beside the ones you already use. Every home in the book comes back with a within-market score, reported as a band inside its own price tier, so the release committee reads it the way it reads turn cost: a reason to release, a reason to hold, weighed against the others.

The construction is the argument. The book is bucketed into price tiers first, then inside each tier the homes are ranked on the score and the weakest are released. Released and retained end up with the same price mix, so a price column has nothing left to sort on, and what the score finds is what remains after price is spent. A release rule sorted on price alone drifts the book down-market while it sells; this one does not.

Same target count, same brokers, same timing. What changes is which addresses are on the list.

Two homes, same NOI, same capex: which one goes?

The one positioned to run behind its market. Comps establish that the two are worth the same today, the operating model says they earn the same, and a release rule built on cost cannot tell them apart. The within-market read ranks each against the homes it competes with, so the committee sees which of the two is positioned to hold its place and which is positioned to slip.

The read is measured on homes already owned for years, scored at a prior sale and observed at the next, which is the seasoned population a holder actually has. It is benchmarked against a price-based release rule on the same book, and the figure that goes on the brief is the one that survives price neutrality, with the raw comparison beside it.

Where does the weak band sit in the book?

A held book cut by metro can look diversified while its weak-band homes cluster in one market or one acquisition vintage. Ranking every home within its own market puts the whole book on one axis, so concentration is measured on something other than location: which markets, which vintages and which price tiers hold the homes positioned to run behind.

Coverage is reported as an output. Homes the model cannot support with enough local evidence come back marked unsupported, with the reason named, and routed to a person instead of ranked on an average. A book whose unsupported homes cluster in one market is telling you where its evidence is thinnest before a single ranked home is examined.

Who runs the review?

Asset management and the portfolio committee at a REIT or a fund holding single-family homes, where the release list is set each quarter and the criteria are operating ones.

Any holder that has stopped adding to the book and is now a net seller, for whom the release list is the only portfolio lever left.

The brief for this seat is built, and it goes out to a desk that asks. The model card and the validation pack go with it on the first call.

What does a pilot look like?

A portfolio with addresses and entry dates. We score every home as of its entry date, using only what existed before it, and return two ranks: the within-market score and the band inside its price tier, keyed on your identifiers with a coverage status per home. The ranks are locked and digested before you release anything, so a rank changed afterwards would stop matching the digest you hold.

You compare the list the score would have released against the list you released, on the homes whose later sales you now hold. This is the one pilot that runs the open test on your own book, and a negative result is a result you keep.