Institutional apartment decks all show the same three-part return bridge. Which line is actually carrying it?

Basis, and it is not close. Take a 148 unit 1998 build in a second tier Sun Belt submarket bought at $151k a door where the same asset traded at $214k in early 2022. Nothing the operator does in the next four years is worth $63k a door of value. An interior scope of $11,500 a door for maybe a $185 rent bump is roughly $28k of value at a 6 cap if every unit turns and holds, which it will not in year one when three lease up comps within two miles are still handing out six weeks free. Where your framing deserves pushback, marlow, is in treating rent recovery as a separate line from basis. The reason that price is available is the same reason rents are soft. You are buying the softness. Model them separately and you have to admit you are double counting the discount somewhere. The line that should never be underwritten is the exit cap. Set the exit 25 basis points wide of entry, and if the deal dies there, it is not a deal.

Quarry, the double counting point is the one I got stuck on when I built this out for myself. If the discount to peak exists because rents are soft and vacancy is 8.5, then buying the discount and then also crediting yourself the rent recovery is one thing counted twice.

Unless the discount is bigger than the softness justifies. That's the actual bet, and I have never seen a deck try to separate the two. What would that even look like, current NOI capped at a normal cap rate versus the price, and the gap is the sentiment discount?

The exit cap option on the poll is mislabeled in most models. Exit cap is a rate bet with a spread bet stapled to it, and those two move independently. Spreads on agency multifamily paper widened and tightened twice through the last cycle without the ten year doing much. The number to watch on the debt side is what the loan constant does to going in coverage. Buy at a 5 in place cap with an all in constant above that and the deal is negative leverage on day one, paying for the recovery in interest while it waits. Some institutional buyers accept that because the equity check is patient and the fund life is long. A lot of the decks marlow is describing show it as a temporary condition that resolves in eighteen months, and that resolution is a forecast, while a forecast is never a term in the loan documents. Confirm the actual rate, the term and any cap requirement in writing with the lender, because what shows in an OM footnote and what shows in the credit approval are frequently different numbers. Basis gets my vote, for the reason quarry gave. It survives being wrong about everything else.

If basis is the whole answer, what stops other buyers bidding it away through 2026 and closing the gap before you deploy?

Halyard, that's my question too. A discount everyone can see and everyone has capital for doesn't stay a discount long.

The value-add line is the one to argue down hardest. Quarry's $11,500 a door is about right for a light interior on a late nineties build, and that same scope ran around $7,200 in 2019. Cabinet fronts, flooring, appliances and the labor to install them all moved. So the renovation premium needed just to stand still is higher than it was, and the rent bump you can prove out with comps is capped by whatever the newest lease-up two miles away is charging net of concessions. Where value-add still works is exteriors and systems on assets where the previous owner deferred. Roofs, parking, amenity spaces and utility billing setups. That work does not show up as a headline rent bump, so brokers underwrite it thin, and it changes the tenant you attract. On the right asset it is worth more than another quartz countertop. Voted basis. Value-add is better understood as what protects the basis rather than what adds to it.

Girder, that last line is doing a lot of work. Value-add as defense rather than upside is not how any deck words it.

Nobody has priced the concession burn-off yet, and it is the piece third party managers on the ground see week to week. In the Sun Belt metros that took heavy 2023 and 2024 deliveries, net effective rents are running eight to eleven percent below face on those vintages. Face rents on renewals are flat to slightly up. So on paper the market looks stable and the collected number is well below it. When the pipeline contraction shows up in deliveries, the first thing that happens is the concession coming off, one week at a time, well before any rent increase. That drops straight to NOI without a single rent roll headline. For a 300 unit asset at $1,650 face, taking six weeks free down to two is roughly $46 a unit a month of real revenue, call it $165k a year, and at a 5.5 cap that is about $3 million of value for doing nothing but existing when supply thins out. I voted rent recovery, and concession burn-off belongs inside it. Marlow, ask your sponsors for net effective rents by vintage in the submarket instead of average asking. If they do not have it, they are reading about the market rather than operating in it.

I voted basis, and cobalt's post nearly moved me. What kept me there is hold length. The hesitation with every one of these funds is the same. The thesis says patient capital acquires at the trough and holds through the recovery, and then the LPA gives you a five year term with two one year extensions at the GP's discretion. If the recovery runs later than the fund life, the manager sells into whatever 2031 looks like, and the LP has no say in it. Basis is the only piece of the return that does not care when you sell. Everything else on marlow's list needs a specific window to be open. Halyard and wren, on the discount closing. It has been closing in the best submarkets already, and the assets still trading at a real gap to replacement are the ones with something wrong, either a submarket with four more years of deliveries or an asset with capital needs nobody wants. So the discount is available, and it comes attached to the reason it exists. The discount still deserves attention, and the work is reading which reason you are being handed. The question to put to the sponsor is what happens if the exit is 2032 instead of 2029, and whether the debt survives that.