WE GRADED OUR OWN UNDERWRITING AND HERE ARE THE RESULTS

Twenty of our loans were paid off and closed out this year, which means twenty times we found out whether we had been right about something nine months earlier.

In lending, being extra right pays nothing. Only being wrong costs anything.

I want to spend this quarter's letter on a claim I think is under-argued, and then show you my own numbers against it.

I. The Asymmetry Nobody Built For

Start with the shape of the instrument.

If you own equity in a project and the property finishes 20% above what anyone projected, you capture that. Your upside is a function of how right you were, and the more right you are, the more you make. Being extra right is the entire game.

Debt does not work that way. A first-position loan that pays a fixed coupon pays exactly that coupon if the house sells for a million dollars, and exactly that coupon if it sells for a million four. Being extra right earns a lender nothing at all.

But being wrong costs everything. If the value comes back materially below what was underwritten, the equity cushion thins, and past a certain point the loan stops being secured in any meaningful sense and becomes an unsecured bet on a borrower's character.

Debt is an asymmetric instrument. Capped on one side, open on the other.

Which produces a conclusion most of the industry talks around: for a lender, forecasting accuracy is not a nice operational quality. It is the only performance metric that exists. Everything else, the coupon, the fee structure, the origination volume, is decided on day one and does not move. The only variable left in the entire transaction is whether the value estimate was right.

And almost nobody publishes theirs.

II. Nine Months of Evidence

So here is mine, sample size attached.

Through the first nine months of 2026, 42 Solutions funded 30 loans totaling $16.6 million, at an average loan size of $553,000. Twenty of those loans have been paid off and closed out. Average life from funding to payoff: 7.5 months against twelve-month terms.

Across those twenty completed projects, our estimated after-repair value averaged within 3% of the price the property actually sold for.

Three additional facts, because a track record without its failures is a brochure.

Zero principal lost since inception. Across 86 loans and $34.6 million.

One foreclosure in the history of the company. February of this year, a $465,000 loan. My wife Camille relisted the property, it sold in thirty days, and we recovered 100% of principal and every dollar of accrued interest.

The book currently sits at roughly 51% of after-repair value. Meaning across every active loan, the property would have to lose about half its finished value before the principal is exposed.

Those figures describe 42 Solutions' lending activity. They are not the performance of any fund, and past results do not predict future ones.

III. Why Most Value Estimates Are Bad

Now the part that generalizes beyond me, because I do not think the accuracy problem is a competence problem. It is an incentive problem.

Ask who produces the valuation and who pays them.

In a great deal of real estate lending, the value opinion is produced by someone compensated per transaction, reviewed by someone who did not visit the property, and approved by a committee measured on deployment volume. Every person in that chain is paid more when the number comes in higher, and no one in that chain finds out whether it was right, because by the time the property sells the loan has been securitized, sold, or forgotten.

That is not corruption. It is drift. And drift compounds quietly, which is exactly why it is dangerous. You do not discover a valuation culture has gone soft in the good years. You discover it in the first bad one, all at once.

Our arrangement is not virtuous, it is just structurally harder to drift in. Camille produces the valuations. She is a Compass agent who lists in these neighborhoods. When a project finishes, she is frequently the one who sells it, which means the person who estimated the value is the same person who later has to face the actual market with it. She finds out whether she was right, on every single file, within months.

That feedback loop is the whole mechanism. There is nothing clever about it.

IV. Grade the Estimator, Not the Estimate

Here is the framework I would hand any allocator looking at any credit manager, private or otherwise. Four questions.

One: Do they publish an error rate, and on what sample? Not a default rate. An error rate. How far off were their value assumptions on completed positions, and across how many. A manager who cannot answer has never measured, and a manager who has never measured is not managing this risk. They are absorbing it.

Two: Who produces the valuation, and who pays that person? Follow the incentive all the way to its end. If the person estimating value is paid more when the number is higher, and never learns whether they were right, you do not have an estimate. You have an advertisement.

Three: How fast does being wrong become visible? This is duration in disguise. On a seven-month loan, a valuation error surfaces in seven months and can be corrected in the next underwriting cycle. On a seven-year hold, the error compounds silently for most of a decade. Short duration is not only a liquidity feature. It is an error-correction feature, and I think that is the more important of the two.

Four: What is between the estimate and the loss? If the value estimate is wrong by 15%, what absorbs it? A manager should be able to answer that in one number without reaching for a spreadsheet.

Four questions. None require you to predict anything about the future. That is intentional. It is the same standard I hold my own underwriting to.

V. What Did Not Go According to Plan

An operating letter that only contains wins is marketing wearing a memo's clothing, so.

Not every project finished on the timeline it was underwritten to. Some ran long. A handful needed extensions, and I granted them.

I want to be precise about how I think about that, because extensions get read as failures and usually are not. When a competent operator is behind on a good property, the alternative to an extension is forcing a sale at whatever moment the calendar happens to land on. That is how a lender converts a timing problem into a value problem, and it is how you lose money on a loan that was never actually in trouble. An extension on a sound asset with a working borrower is the system functioning, not failing.

An extension on a project that has gone quiet is something else entirely, and those get handled differently and immediately.

The one file that did not recover went to foreclosure in February, and I have written about it before. The lesson from it is now a permanent rule here: twelve months of borrower payment reserves, verified, before funding. That rule exists because a deal taught it to me, which is the only way rules like that ever get made.

VI. The Part for You

The reason I wrote this quarter's letter about measurement rather than about markets is that the question generalizes past lending entirely.

Everywhere you have capital, someone made a forecast. A sponsor projected an exit price. An operator projected a lease-up. A manager projected a growth rate. You are being paid, in part, for the accuracy of forecasts that other people made and that almost none of them ever go back and grade.

Take your three largest positions. For each one, find the specific forecast the investment rests on, the exit cap rate, the ARV, the terminal value, whatever the load-bearing number happens to be. Then ask one question: has the person who made that forecast ever told you how accurate their last twenty were?

If the answer is no on all three, that is not a reason to sell anything. It is a reason to ask on your next call, and to notice very carefully how the question is received. A manager who measures will have the number close at hand and will usually be glad you asked. A manager who does not will tell you why the question is more complicated than it sounds.

Both answers are information. Reply and tell me: on your largest current position, who made the load-bearing forecast and have they ever graded it?

P.S. This is the third C doing its job quietly. Character, Capacity, Collateral, Capital. Collateral is the only one of the four that still protects you when you were wrong about the other three, and collateral is worth exactly what somebody estimated it to be worth. Which is why the estimate is not a formality. It is the foundation the other three stand on.

The lending figures above reflect 42 Solutions, LLC's lending activity as of September 30, 2026. They are not the performance of any fund. Past performance is not indicative of future results. This letter is educational and is not an offer to sell or a solicitation of an offer to buy any security. Real estate lending involves risk, including loss of principal.

Newsletter Edition #213