A bulk trade is not a harder version of a flow sale. It is a different exercise, with a different failure mode: flow sales fail on pricing, bulk sales mostly fail on readiness. The bid stage is short. Everything before and after it is where the calendar goes, and nearly all of it is under the seller's control.
What follows is the sequence we see work, written from the seller's side.
Stage 1: Decide what the pool is for
Before anything else, be specific about the objective, because it determines the pool. Freeing warehouse capacity, exiting a product line, rebalancing concentration, and monetising a seasoned book are four different goals and they build four different pools.
The most common early mistake is assembling the pool from what is easiest to extract rather than from what serves the goal. That produces a lot that is hard to describe, which makes it hard to price, which slows everything downstream.
A note on carve-outs
Deciding in advance which loans you will not sell at any price is worth an hour of your time. Buyers re-cut pools constantly and it is far cheaper to state a restriction up front than to withdraw loans after a bid is out.
Stage 2: Build the tape before you start conversations
The tape is the trade's shared reference. Every downstream step reads from it, and the failure mode is not a wrong tape; wrong is caught. It is an ambiguous tape, where two readers reasonably disagree about what a column means.
- One row per loan, one header row, no merged cells, no subtotals inside the data.
- Every balance on a single stated as-of date, stamped on the file.
- A data dictionary alongside it. Two lines per column beats a phone call about column AH.
- Blanks left blank. A zero where the value is unknown is worse than an empty cell, because a zero is read as information.
- Consistent identifiers you can carry all the way through settlement, so the loan numbered 4417 is the same loan on the tape, in the data room and on the final schedule.
If you want the field-level detail on what a tape should contain, that is its own article: see What Makes a Loan Tape Price Well.
Stage 3: Assemble the data room in parallel, not after the bid
This is the highest-leverage item on the list. Diligence requests are predictable, so preparing for them is not speculative work. It is work you will certainly do, moved earlier, when it is not on the critical path.
- Underwriting guidelines in force at origination, by vintage. Current guidelines do not describe a loan written eighteen months ago.
- Your QC and audit results for the population, including exceptions and how they were cured.
- Servicing records: the full pay string, escrow status, any loss-mitigation history.
- Valuation documentation, and any subsequent valuations.
- Licensing and compliance testing evidence for the states represented.
- A sample of complete loan files, chosen to be representative rather than flattering. Buyers notice curated samples, and it costs credibility that is expensive to rebuild mid-trade.
Stage 4: The bid stage
With a clean tape, indicative pricing is fast: our desk works in days, not weeks, and the models read every loan in the file rather than a sample. Two things are worth understanding about what you receive.
First, an indicative bid is exactly that: indicative, subject to diligence and final documentation. It is a real number produced by real analysis, not an anchor, but it is conditioned on the tape being an accurate description of the collateral.
Second, the number is not raw model output. A trader reviews it before it is released. If a segment prices unusually, that is a conversation worth having rather than a line item to accept or reject. Often the explanation is a data question rather than a credit view, and data questions are cheap to fix.
Stage 5: Diligence without drift
Diligence goes wrong in a specific way: the pool moves while it is being examined. Loans pay off, borrowers go delinquent, files get withdrawn, and the schedule under review stops matching the schedule being priced.
- Fix a cut-off date and hold every party to it.
- Name one person on each side who owns the loan schedule. Two people maintaining versions is how a settlement slips a week.
- Report material changes as they happen rather than in a batch at the end. A payoff disclosed on the day it occurs is administrative; the same payoff disclosed at settlement is a re-cut.
- Expect kick-outs and agree the mechanism for them in advance. Every bulk trade has some. What varies is whether the process for handling them was defined before or after the first one appeared.
Stage 6: Settlement mechanics
The last stretch is logistics, and it is boring in the best case. Confirm the final loan schedule against the final purchase price and settle the accrued-interest and escrow calculations early. Sequence the collateral file delivery and the custodial certification; a certification exception found after funding is a slow problem. Line up the servicing transfer date and the borrower notification requirements, and if servicing is being retained, make sure that is documented in the agreement rather than assumed from the conversation.
What actually compresses the calendar
Nothing in this list is exotic. The trades that settle quickly are not the ones with the simplest collateral. They are the ones where the tape was unambiguous on day one, the data room was populated before it was requested, and one named person on each side owned the loan schedule from cut-off to funding. That is most of the difference between a bulk sale that takes weeks and one that takes months, and effectively all of it is decided before a buyer sees the first file.
