Talked myself out of a deal for 14 months and then the market moved. What it cost

The line about building models until one said no is going in my notes. I've done exactly that on smaller purchases and never had a word for it.

One thing I'd note in your defense though. The stress case that broke even was a 15 percent overrun plus six months of extra lease-up, and that isn't an absurd combination on a first ground-up with a GC whose experience was renovation. Break-even in a bad case isn't a disaster outcome, it's actually a survivable one, which is the reading you didn't take.

@cairn has it. Break-even in the double stress case is a pass, not a fail. The test most people should use is whether the bad case wipes out the equity, and yours didn't.

The other thing in the post that I'd flag is the term sheet. You had 68 percent loan to cost on a first-time ground-up sponsor with a renovation GC. That's a lender telling you they believed the project. Lenders at that stage are usually more conservative than the borrower, and when they're not, that's information.

Genuine question, how do you write down a go/no-go test in advance without it just being a number you move later? What stops you from deciding the threshold was wrong once you see the answer?

@plumb the version I've seen described is you write it down and give it to someone else, a partner or an investor, before you run the numbers. Then moving it is a conversation instead of a private edit. Doesn't fully solve it but it raises the cost of moving.

I don't understand the yield on cost thing well enough to follow whether 6.6 was actually good. Good compared to what?

@vellum, yield on cost is your stabilized NOI divided by everything you spent to create the building. You compare it to the cap rate you'd pay to buy a similar finished building. The gap between them is your development spread. If finished buildings in that submarket trade at, say, a 5.3, then building at 6.6 means you created something worth meaningfully more than it cost. That spread is the entire reason to develop instead of buy.

130 basis points is on the thin side of what most people want, and it was probably enough in a pocket with no competing supply.

The 68 percent loan to cost is doing a lot of work in this thread and I want to add one thing. Term sheets aren't commitments. Between the term sheet and closing there's appraisal, there's the lender's own construction cost review, and there's whatever their credit committee does with a first-time ground-up sponsor. @windlass, did you ever get to a commitment letter, or was the 68 percent still a proposal when you walked?

I ask because a proposal at 68 that becomes a commitment at 60 changes the equity requirement by about $179k on a $2.24M project, roughly $717k in the deal instead of $897k, and that's a real reason to hesitate.

The GC pricing at $1.71M on a renovation background is the piece I'd have worried about most, coming from the trades. A guy who's good at gut rehabs is not automatically good at buying out a ground-up wood frame job. The subs are different, the sequencing is different, and the number he gives you is usually assembled from fewer real bids than you think.

Did you ever see the buyout, meaning actual sub quotes behind the $1.71M, or was it a lump sum with a schedule of values?

What strikes me is that the buyer paid $25k more for the land and built the same plan. So the deal wasn't marginal on price, it was marginal on your confidence. That's a different problem and it doesn't get fixed by more spreadsheet work.

I'd want the boring version of the counterfactual before anyone calls this a loss. The buyer built it and it's leasing. We don't know their cost overruns, we don't know their carry, and we don't know what their equity return looks like. Leasing at 6 percent above @windlass's rents is one number and it doesn't tell you the project made money.

@verity is right to be careful but rents 6 percent over underwriting on a project that penciled at 6.6 is a lot of cushion. That's roughly 40 basis points of yield on cost handed to them for free.

The thing I'd add from the renovation side, @windlass, is that your 15 percent overrun case was probably the right stress number but you applied it to the wrong line. On ground-up, hard cost overruns tend to cluster in sitework and utilities, which you can partially price out before closing with the geotech you already paid for. Interior finish overruns are the renovation problem and they're much less of a ground-up problem because you're building to a fixed drawing. Splitting the contingency by category would have given you a less scary and more accurate stress case.

Worth saying that $20k of dead deal cost is the price of the option to walk, and it's a reasonable price. People treat abandoned pursuit costs as pure waste. Every developer I've talked to budgets a pursuit cost line knowing some percentage of it dies.

@windlass, the part that would bother me more than the deal is what the 14 months of re-running did to your pipeline. You weren't looking at other sites during that, were you? The real cost of analysis paralysis is usually the deals you never saw because you were busy with the one you didn't do.

The fix I use now is a clock. Two weeks of feasibility work, then a decision, and if I can't decide in two weeks the answer is no and I move on. It means I say no to some good deals. It also means I've looked at four times as many.

I've done zero deals, and what stuck with me here isn't the $20k. It's that you knew the spread was fine at the time and passed anyway. I've been assuming my problem will be not knowing enough. Sounds like the problem is doing something with what you already know.