Where to Point the Lens First

A CMBS book is thousands of loans; you have attention for dozens. So the first skill in surveillance is not reading a loan — it is deciding which loans to read. This chapter is about triage: spending cheap signals to spend your expensive attention well, and reading the disagreement between two screens as a map of where to look.

Reading a loan properly — standing it against its maturity, its structure, its stack — is expensive. Your attention is the scarce resource, and there are far more loans than you can ever spend it on. The temptation is to start at the top of the loan list and grind, but that is exactly backwards: you will burn your best hours on healthy loans and arrive at the dangerous one too tired to see it. The discipline is coarse-to-fine. Spend a cheap, mechanical signal across the whole book first to decide where the expensive reading goes. Triage is not a shortcut around the reads that fill the rest of this course; it is what points them.

You have two triage instruments, and they are usefully different. The first is the Stress view — the platform’s own ranked screen, one number per loan built from coverage, leverage, and the exit, so you can sort thousands of loans and simply read from the top. It is your model’s opinion, applied uniformly, unafraid to rank a loan the market still likes. The second is the Watchlist — the servicer’s own flags, the loans the people physically closest to the collateral have already put on notice. One screen is a model’s judgment; the other is a disclosure from the party holding the keys. Neither is truth, and that is the point.

Because the real signal is in the disagreement. A loan that lights up on both screens is a loud, redundant “look here” — start there, but expect few surprises; everyone can see it. The information is in the loans where the two lists diverge. A loan high on your stress score but absent from the watchlist is the most interesting object in the book: either you have caught a deterioration the servicer hasn’t disclosed yet, or the servicer is late, or your model is wrong — and finding out which is exactly the kind of lead that surveillance exists to chase. A loan on the watchlist but low on your score is the mirror: a servicer being conservative, or a signal your model doesn’t yet capture. You are not looking for the list that is “right.” You are reading the gap between them as a map of where your attention is most likely to be repaid.

Those two screens find where distress is. A third finds where it is going. The Transfer Risk panel ranks still-performing loans by the modeled probability they transfer to special servicing within roughly the next two quarters — the pre-emptive triage. Point your lens at the top of that list and you are reading distress before it is distress, which is the only place an outside reader can actually add value: after the transfer, the market can see it too, and the price has already moved. Stress and Watchlist tell you what is broken; Transfer Risk tells you what is about to.

Hold triage in its proper place, though: it generates hypotheses, it does not settle them. A ranked screen with a huge “distressed” tail is almost always an artifact — the same lesson that runs through every part of this platform, that a too-large signal is a measurement error until the known count says otherwise. A screen that agrees with the watchlist perfectly is probably just re-finding the servicer’s disclosure, not adding to it. Triage hands you a reading list of a few dozen names; the reads in the chapters ahead are how you adjudicate them. Point the lens well, and the rest of the course is you finally looking through it.