CLO compliance that traces every trustee report break to the loan before the payment date
Every break traced to the loan before the payment date, not after the report is final.
See one case, screen by screen ↓CLO compliance is the manager's monthly check that a CLO passes its coverage, collateral quality and portfolio profile tests as its indenture defines them, by reconciling the trustee's report with the manager's own numbers. CLO Compliance gives that work to seven agents: they read each trustee report, recompute every test from the manager's own positions, and explain each break to the loan before the payment date. An analyst confirms each cause and the head of CLO operations signs.
AI in asset management: a buyer's guide to agentic AI for investment managers →
Updated
CLO compliance comes down to one monthly question: are the trustee's numbers right? Each trustee report sets the coverage tests that decide where the interest goes, the collateral quality tests that limit what the manager may buy and the portfolio profile limits it must respect. The manager has about eight days to reconcile it with its own books and its own reading of each indenture. Today that means re-keying the report, recomputing the tests in a coded compliance engine or a shadow trustee's file, and tying out test by test, then loan by loan, in spreadsheets.
The tools in that chain list differences, not causes. A 46 bp gap on a Class D OC test can come down to one loan treated under one definition, but finding it means reading the indenture, the trade ticket, the price history and the asset schedule. When the cause turns up after the trustee finalizes, a test that should fail has already passed, and on the payment date the waterfall pays the equity money the indenture says must buy collateral. The fix waits for the next period, or never comes.
Estimated split of the eight-day window for one analyst's book of about 16 CLOs, by hand and with the solution.
Seven agents read, recompute, explain and check every test; an analyst confirms, the head of CLO operations signs.
Quillfield Capital manages 31 CLOs, 22 US and 9 European, with $14.8 billion of loans in 12,184 positions. On Monday, October 12, the draft payment date report for Quillfield CLO 2023-1 arrives from Ashgrove Corporate Trust. Ashgrove finalizes on Thursday; the deal pays on Wednesday, October 21. Here is what happens next, screen by screen, in the working solution.
Priya's home shows 28 of 31 trustee reports in, 34 breaks explained as 17 causes, 10 trustee queries and 9 days to the next payment date where a break moves cash. Beside it, each agent says what it did overnight, every line cited: the Reader agent read 28 reports (1,755 pages), the Recompute agent recomputed 1,766 tests from 12,184 positions, and the Diff agent grouped the breaks. She can ask or instruct the agents in plain words from the same panel.
Payment-date watch agent: “What moves cash first: in Quillfield CLO 2023-1 the interest diversion test fails on our numbers, so $583,750 must buy collateral on October 21.”
The deal opens on a row of test gauges. Each carries the trustee's mark, Quillfield's own mark recomputed from the indenture, and the trigger between pass and fail. Class D OC reads 105.08% for the trustee and 104.62% on Quillfield's numbers, against a 104.10% trigger. The interest diversion test passes for the trustee and fails by 13 bp on Quillfield's. Of 62 tests, 54 agree; 8 breaks come from 3 causes, with a countdown to Ashgrove's Oct 15 finalization and the Oct 21 payment date.
“62 tests · 54 agree · 8 breaks from 3 causes.”
The Diff agent rebuilds the Adjusted Collateral Principal Amount both ways, line by line. Two lines move together: the trustee keeps $9,680,000 of the Alder Park Holdings term loan B in the principal balance at par, with no discount flag on its asset schedule (page 19), while Quillfield, which bought it at 78.50 on March 18, 2026, carries it as a discount obligation at purchase price. Ashgrove flagged it as a discount obligation through September; its October file dropped the flag. The net effect is $2,081,200 of adjusted par.
Checker agent: agrees · 97%, with the seven sources it read, each cited.
Every figure has a citation that opens its source. The definition of Discount Obligation (§1.1, page 47) opens with the passage marked: a loan bought below 80.0% of its principal balance stays a discount obligation until its market value has been at least 90.0% for 30 consecutive business days. Alder Park's high since purchase is 86.50, on September 4, so it has none of the 30 days it needs. It is still a discount obligation.
The same cause moves five tests: Class A, B, C and D OC and the interest diversion test. The OC tests still pass on Quillfield's numbers, but the interest diversion test does not: 104.62% against 104.75%, while the trustee's report shows a pass. On October 21, the lesser of 50% of the remaining interest ($958,700) and the cure amount, $583,750, must buy collateral. The subordinated notes receive $897,550 instead of the $1,481,300 the trustee's numbers would pay.
“Fails · trustee shows a pass.”
The agents explain; Priya decides. She confirms cause 1 with the reason code “Trustee error — definition not applied” and a note for the record. The other reason codes sit beside it: stale data at the determination date, a trade or paydown in flight, a fix in Quillfield's own system, an interpretation ruling by the head of CLO operations, or within tolerance.
“Recorded as Priya Raman · CLO Compliance Analyst · Mon Oct 12, 8:40 a.m. · the audit trail keeps the evidence the agents cited.”
The Query agent has drafted TQ-1031 to Ashgrove's CLO reporting desk with the evidence attached: indenture pages 47, 31 and 58, the trustee's own page 19, the trade confirmation and the price history. Priya submits it at 8:49 a.m. Marcus Feld approves it as the checker, with the note “Evidence checked: definition p. 47, trade ticket, price history.” Daniel Okafor, the trustee relationship manager, sends it at 8:57 a.m.
“Agents never send. The maker and the checker must be different people; the platform enforces it.”
On the breaks screen Priya tells the Diff agent to accept the three timing and data differences. It finds them: Kinlock Pharma settled after the trustee's cut-off, Orchard Valley Foods' $3.1M paydown is not yet in Quillfield's holdings, and Ashby Leisure was converted a day apart. It selects all three, seven breaks in all, for her to accept with one reason code, and says it will check the next report for the reversal.
Diff agent: “I have selected them; accept them with one reason code, and I will check the next report for the reversal.”
On Quillfield CLO 2020-1, Redfern Retail missed its September 15 coupon and is a defaulted obligation that the trustee still carries at par. The Diff agent values it at the evaluated mid of 39.25 (Class D OC 104.67%); the Checker agent reads “bid-side prices” as the bid of 38.00 (104.66%). Both readings stay above the 104.48% trigger. Because the agents disagree, nothing is called explained: Priya chooses the price side, and only then is the query to Calloway drafted.
“Which price side should the query to Calloway use?”
Ashgrove agrees and reissues the report: Class D OC 104.62%, the interest diversion test failing and $583,750 diverted. The Reader agent ties out the revised report, and after the Halvorsen Marine industry fix, which Marcus approved as the checker, 62 of 62 tests agree. The Evidence agent's checklist is complete. Priya prepares the sign-off, and Marcus signs it with his second factor at 9:11 a.m., nine days before October 21.
“Maker ≠ checker · enforced by role.”
The same trustee report, seen by the five people who carry it: what their cycle looked like, and what it looks like now.
Seven agents read, recompute, explain and check every test; an analyst confirms, the head of CLO operations signs.
Picks up each trustee report from the portal, email or SFTP with its data file, extracts every test, trigger, asset row, trade and cash line with page citations, and checks the totals (sum of par against the reported aggregate). It also reads trustee replies, revised reports and agent notices in their own language.
Recalculates every test from Quillfield's own positions as of the determination date, using the deal's approved indenture definitions, to give the manager's second mark.
Explains every break down to the loan, groups breaks into causes, measures the effect on each test and classifies the cause as a trustee error, Quillfield data, timing or interpretation, with a proposed action.
Re-derives each explanation on its own, from an asset-level rebuild on a different model family, and agrees or challenges.
Drafts the trustee query with its evidence attached, tracks replies against the two-business-day reply time, drafts chases, reads the answer and reopens or closes the break.
Keeps every deal's clock (determination date, trustee finalization, payment date), runs the priority of payments on both sets of numbers and escalates at T-10, T-5 and T-2.
Builds each deal's sign-off checklist from the audit trail and assembles the month's evidence pack: the report, recompute workpapers, break log, queries, sign-offs and their hashes.
Maker, checker, sender. The agents propose; a named person decides.
Ask in plain words, or tell them what to do. Every answer cites its page or record, and the agents open the screen it is about.
Why does Class D OC differ in Quillfield CLO 2023-1?
The trustee carries Alder Park Holdings at par. It was bought at 78.50 on March 18, which makes it a discount obligation carried at purchase price, and it has never traded at 90 or above. That removes $2,081,200 of adjusted par: 104.62% on our numbers against the trustee's 105.08%, trigger 104.10%. The interest diversion test fails on ours, so $583,750 buys collateral on October 21.
What is the CCC bucket in 2023-1?
7.31% on our numbers and 6.84% on the trustee's, against a maximum of 7.50%: it passes with 19 bp of cushion on ours. The two marks are 47 bp apart because of Castellan Foods and Brightwater Dental: the trustee's rating file dated October 1 misses two downgrades.
What if we sell Alder Park at 84.25?
The sale alone does not cure the interest diversion test. Class D OC rises from 104.62% to 104.74%, 1 bp short, so $27,150 of interest would still be diverted on October 21 instead of $583,750. Nothing was traded; the what-if is saved for Elena Brandt.
Which CLOs are closest to a coverage trigger?
On our numbers: 2020-1 is 18 bp over its Class D trigger (Redfern Retail open), 2023-1 is 52 bp over but below its interest diversion level, and 2019-2 is 131 bp over. The CLOs are sorted by cushion.
13 screens from the working solution, on its sample data. Pick one to see it large.
When a difference counts as a break, who must look again and when the clocks run are settings. The administrator proposes a change with a reason, and it takes effect only when the head of CLO operations approves it.
| Setting | Default | Choose from |
|---|---|---|
| Coverage tests (OC, IC) | 1 bp tolerance | Second look over 25 bp |
| Interest diversion / reinvestment test | 1 bp tolerance | Second look on any difference that changes the result |
| Max WARF | 5 | Second look over 25 |
| Min weighted average spread | 1 bp | Second look over 5 bp |
| Position price and par | 0.125 · $1 | Differences below these are not breaks |
| Trustee reply to a query | 2 business days | An open coverage break escalates to the head of CLO operations 3 business days before the trustee finalizes |
| Payment-date alerts | T-10, T-5 and T-2 | For any deal with an open break |
| Retention | 7 years | Advisers Act Rule 204-2 requires 5; legal hold can be placed per deal |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample data · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (Wilmington Trust, CLO reporting needs, 2023, on the typical eight-day reporting window; PitchBook via WorkWise, US CLO issuance about $203 billion in 2025 and over $600 billion of broadly syndicated CLOs outstanding). People, firms, trustees, deals and borrowers named on this page are fictional: characters and sample data in the working solution.
CLO compliance is the manager's monthly check that a CLO passes its coverage, collateral quality and portfolio profile tests as its indenture defines them, by reconciling the trustee's report with the manager's own numbers. CLO Compliance gives that work to seven agents: they read each trustee report, recompute every test from the manager's own positions, and explain each break to the loan before the payment date. An analyst confirms each cause and the head of CLO operations signs.
They are the tests each indenture sets: coverage tests (overcollateralization and interest coverage by class, and the interest diversion test), collateral quality tests (maximum WARF, minimum weighted average spread, minimum diversity) and portfolio profile limits such as the CCC bucket. In the working solution, Quillfield CLO 2023-1 has 62 of them, each shown with the trustee's mark, the manager's mark and the trigger.
The overcollateralization (OC) test divides the Adjusted Collateral Principal Amount, as the indenture defines it, by the notes outstanding at that class and above, and compares the result with the trigger. The Recompute agent rebuilds it from the manager's positions with the indenture's own definitions: on Quillfield CLO 2023-1, $473,410,000 against $452.5 million of Class A–D notes gives a Class D OC of 104.62%, against the trustee's 105.08%. The Diff agent then traces the difference, line by line, to the loan behind it.
It is a coverage test, often set above the junior OC trigger, that diverts interest away from the equity to buy collateral when it fails. The mechanics differ by deal, so the solution reads them from the indenture and cites the page. In the sample deal it fails on the manager's numbers at 104.62% against 104.75%, so $583,750 must buy collateral on the payment date.
It gives the manager its own second mark, like a shadow trustee, but it does not stop at a third set of numbers. The agents explain each break to the loan, group breaks by cause, show what each does to the payment-date waterfall, and draft the trustee query. The trustee's report stays the record the solution ties out.
Yes. The working solution runs 9 European CLOs alongside 22 US deals. The Reader agent reads agent notices in German, French, Dutch and Spanish and shows the original beside the English, with each figure pointing back to the original line. European deal data is stored and processed in the EU (Ireland).
Yes. The agents explain, check and draft; they never send a query, sign a report, change a definition or decide an interpretation. The analyst confirms or rejects each cause with a reason code, a second person approves every query, fix and sign-off, and the head of CLO operations signs with a second factor.
The Agentic Solution Engine builds and deploys it from your requirements and documents: your indentures, a sample of past trustee reports and your tolerances. It goes live once every quality gate has passed. We will walk you through it on your own reports first.
Every credit question answered in minutes, each sentence pinned to its tranche and page.
1-min demoBond Covenant AnalysisEvery basket in dollars and every leakage route found, before books close.
1-min demoProspectus ReviewEvery deal that fits a fund briefed and decided before books close, every line cited.
1-min demoWe'll run CLO Compliance on a sample of your own past trustee reports and indentures.