Treating every deed conversion as a failure changes how a lien model has to be built

That 0.5 percent holds because the fund hired collectors instead of acquisition analysts.

The part that would wreck my model is redeployment speed, not yield on any single cert.

Florida's 2-year redemption window and NJ's 6-percent statutory penalty basically bribe owners into paying, so 0.5 percent is easier there than it sounds.

The framing of "collections vs. real estate" is useful, but it undersells how much the 7.1 net depends on the specific municipal profiles they picked, beyond the operational posture. Florida and New Jersey both have dense suburban tax bases where owners almost always have equity and almost always redeem, so a collections-first approach works because the underlying population of certificates supports it. Take that same philosophy into rural Georgia or Mississippi, where a meaningful slice of the book is vacant land with no equity and no owner who will ever answer the phone, and the 0.5 percent holds zero. Before rebuilding a model around his numbers, the question to ask is what percentage of his pool by count, rather than by dollar, is residential homestead.

The skill gap between a good skip tracer and a good title examiner is about 40 years of compounding problems.

The guy you talked to is not wrong but I would push back on treating deed conversion as failure in every case. That framing works when your pool is 90 percent residential owner-occupied in zip codes like 33712 or 07202 where someone almost always redeems because they have equity and they care about the house. We have certificates right now in a pool we started building in late 2022, mostly Duval County and a pocket of Cumberland County NJ, and the parcels where we ended up taking deed or getting close to it were not failures in the underwriting, they were the thesis. Vacant lots adjacent to assemblage plays, a small commercial strip where the owner had been dead for six years and the heirs were three states away and completely unreachable. You are not going to phone-plan your way to redemption on a parcel where there is no living motivated owner on the other end. His 0.5 percent probably reflects a pool composition decision more than a collections operation superiority, and if you built a pool without those parcels you would also have a low conversion rate.

Where I think your model actually has a gap is not collections versus real estate, it is parcel triage speed. The eight-month mark is where we decide which certificates get the outreach budget and which ones get flagged for deed track, and that decision has to be made on something other than time elapsed. We look at equity cushion relative to our lien position, probate status, and whether there is an active mortgage behind us. A certificate in Broward on a property with 200k equity and a first mortgage gets every phone call we can make. A certificate on a tax-delinquent vacant lot in a rural county with no mortgage and no heir engagement gets put on deed track by month ten, not month 24. Waiting 24 months on the second type just

Eight months in on the capital side and you're already catching the model problem. That's actually the right moment to catch it. The 4 to 6 percent deed assumption is not crazy, it's just an assumption about operator behavior, not lien behavior, and those are different things. A fund that runs a collections desk is going to suppress conversion because they're paying for that outcome in labor and systems, and if you're allocating to a GP who doesn't have that infrastructure, your 0.5 percent benchmark is basically meaningless for them. 7.1 net to LP is solid but I'd want to know what their servicer costs look like, because the phone calls and payment plans and skip tracing on a 28 million book aren't free and that number has to be net of something.

The piece worth pushing back on slightly is the assumption that 7.1 net is clean across both states, because Florida and New Jersey have very different carrying costs once a certificate starts aging. Florida OID rules and the way interest resets at redemption are fairly friendly to a patient holder, while New Jersey has municipal charges that can accrue against the liened property and complicate a lien position when nobody is watching the junior certificate calendar closely. A fund running a blended pool across both and quoting one net number is either very good at its internal accounting or it is smoothing something. The collections framing is real, and on the title side that 0.5 percent conversion number almost certainly includes parcels sold or assigned off the book before any deed was taken, because those never show up in a conversion rate. Selling a certificate to a deed-hungry buyer at month 20 is neither a redemption nor a conversion, and it still clears the book. Eight months in on the allocation side is early for that transaction type to have surfaced, and it shows up in the footnotes if anywhere.