Weighted Average Rating Factor (WARF)

By Ryan O'Connell, CFA, FRM Published November 14, 2026

A weighted average rating factor (WARF) summarizes the credit quality of a collateral pool by
converting rating categories into numerical factors and averaging those factors using specified exposure weights.
For an analyst reviewing a collateralized loan obligation (CLO), the calculation helps answer a narrow question:
how does the pool’s rating mix compare with a disclosed credit-quality limit?

The answer depends on the factor table and the transaction’s definitions. Rating letters are ordered categories;
adjacent categories do not represent equal increments of default risk. This guide uses a dated U.S. financing
agreement to demonstrate the arithmetic, then separates collateral quality from diversification, default
probability and the rating of a CLO tranche.

What Does WARF Measure?

Credit ratings express relative creditworthiness. They are judgments based on an agency’s methods and assumptions,
rather than measurements on an equally spaced ruler. The SEC also cautions that rating scales differ among
agencies and that ratings do not guarantee repayment.^[4] A factor table
supplies a particular numerical mapping; the analyst must identify which mapping applies.

CLO collateral commonly includes leveraged loans. The loans’ credit quality and
the protections available to a senior CLO note are different analytical questions. For the structure and payment
priorities, see collateralized debt obligations.

A Dated Factor Table: Identify the Input First

The example uses Barn Owl Funding LLC’s March 25, 2025 credit agreement, filed as SEC Exhibit
10.1. It finances assets in anticipation of CLO securities; it is a contract, not an agency methodology. Schedule
F contains its Moody’s Rating Factor table. The selected entries below are specific to that dated document, not
a statement of current universal Moody’s policy.^[1]

The lookup input is the agreement’s defined Moody’s Default Probability Rating. Schedule E-1
starts with the obligor’s corporate family rating when available, then provides other routes, including adjusted
debt ratings and derived ratings. A displayed facility rating therefore cannot automatically be pasted into the
lookup. This contractual label also should not be assumed to mean an agency-issued probability-of-default rating
bearing a -PD suffix.^[1]

The increasing gaps matter. Ba2 to B1 and B3 to Caa2 each span two rating subcategories, but the corresponding
factor increases are 2,220 − 1,350 = 870 and
6,500 − 3,490 = 3,010. Equal notch movements can produce different factor changes.

The Weighted Average Rating Factor Formula

First define the included set of assets, I. Let P_i be each included asset’s specified principal
balance and F_i its lookup factor. Its weight is its principal divided by total included principal:

In this agreement, Schedule F excludes Defaulted Obligations from both sides of that calculation and rounds the
final average up to the nearest whole number. Retain full precision through the calculation;
ordinary nearest-integer rounding can give a different answer. Use the contractual Principal Balance definition,
rather than switching to market value because a loan trades below par.^[1]

Excluding a defaulted asset does not erase its economic loss. It changes the population measured by this particular
statistic. Report defaults separately, and do not compare two WARFs before reconciling their asset coverage,
rating dates, adjustments and weighting conventions.

Worked Example: One Position Deteriorates

Consider a hypothetical $100 million pool of three fully funded, nondefaulted loans with principal
balances of $50 million, $30 million and $20 million. Assume their contract-defined lookup ratings are Ba2, B2
and B3, respectively. There are no ratings on review, cash balances or other assets, and principal stays fixed.
These assumptions isolate the calculation; this is not Barn Owl’s actual portfolio or a demonstration of full
transaction eligibility.

Now worsen only the $20 million position from B3 through Caa1 to Caa2. Its factor becomes 6,500, and its weighted
contribution becomes 0.20 × 6,500 = 1,300. The revised WARF is
675 + 816 + 1,300 = 2,791. The increase is
0.20 × (6,500 − 3,490) = 602 factor points.

For comparison, a separate two-subcategory deterioration from Ba2 to B1 affecting a 20% position would add
0.20 × 870 = 174 points. The same exposure and notch count produce different sensitivity
because the lookup is nonlinear. Neither change is a percentage increase in the pool’s default probability.

Weighted-Average Ratings vs. Rating Factors

A report labelled “weighted-average credit rating” requires its own methodology. An analyst might assign numerical
positions to rating letters and average those positions, but equal numerical spacing is an added convention.
It does not establish equal default-risk spacing.

To demonstrate, invent a teaching code: Ba2 = 1, Ba3 = 2, B1 = 3, B2 = 4, B3 = 5, Caa1 = 6
and Caa2 = 7. This is not a recognized agency calculation. For the worked pool, its average would move from
0.50 × 1 + 0.30 × 4 + 0.20 × 5 = 2.70 to
0.50 × 1 + 0.30 × 4 + 0.20 × 7 = 3.10.

That invented score rises by 0.40 whenever a 20% position deteriorates two steps, regardless of its starting
category. The factor calculation preserves the unequal spacing in the specified table. Do not round either
average back into rating letters without an explicitly authorized reporting method; a numerical average does
not create an agency rating.

Both measures describe a rating mix at a particular time. Credit risk
migration matrices answer a different question: how ratings can change between periods.

Collateral Quality vs. Diversification

Reset to two different hypothetical $100 million portfolios. Each contains ten $10 million loans,
all with a B2 lookup rating and factor 2,720, with no defaults or review adjustments. The spread portfolio has ten
unrelated issuers across five industries, two issuers per industry. The concentrated portfolio has five loans to
one issuer, three to another and two to a third, all in the same industry.

Both averages equal 1.00 × 2,720 = 2,720. The factor arithmetic has no input for how many
loans share a borrower or industry. More loan records therefore do not necessarily provide more diversification.

For a real contractual example of that separate task, Bluejay Funding LLC’s February 18, 2025 agreement calculates
its Diversity Score using issuer balances, capped equivalent units and industry-level lookup scores. Its minimum
score test is not a raw loan-count test. These are Bluejay’s provisions, not an additional Barn Owl covenant or
a diversity score calculated for the hypothetical portfolios.^[2]

Even different issuers and industries can suffer together. Default
correlation concerns that dependence; a favorable average factor cannot establish that defaults are
independent. Broader credit portfolio models examine loss outcomes
beyond the rating average.

Maximum-Factor vs. Minimum-Diversity Tests

A maximum-factor test puts an upper bound on the specified quality measure. A minimum-diversity
test puts a lower bound on a separately defined diversification measure. Improving one does not
automatically satisfy the other.

In the Barn Owl agreement’s pre-CLO-pricing framework, its Moody’s Maximum Rating Factor Test is satisfied before
the first date aggregate portfolio principal reaches $75 million on a trade-date basis. From that first crossing
onward, its adjusted WARF must be no greater than 3,100. The adjusted definition treats applicable
ratings under review for upgrade or downgrade as moving one subcategory in the corresponding direction.
Our example assumes no review status; an actual report must distinguish adjusted from ordinary
WARF.^[1]

Those terms cannot be transferred wholesale to another CLO. The agreement also changes its collateral-quality
framework at CLO pricing by referring to the Reference Document, and its purchase conditions include
maintained-or-improved tests and lender-waiver provisions. Read the operative document and amendments before
deciding whether a trade is permitted or what a failure does. A standalone WARF comparison is not a full
compliance conclusion.^[1]

How to Check a Reported WARF

1. Bind the definition and date. Identify the agreement, amendments, calculation date and exact
   ordinary or adjusted measure. Confirm the factor table’s version and applicable test level.
2. Resolve each rating input. Follow the hierarchy for issuer, facility, derived or estimated
   ratings. Apply review-status rules before the lookup when required.
3. Reconcile asset coverage. Check principal definitions, default exclusions, unrated assets,
   cash, unsettled trades and commitments. Use the same included assets in numerator and denominator.
4. Recalculate and round last. Verify weights sum to one, check each contribution and apply the
   stated rounding convention to the aggregate.
5. Read the companion measures. Review defaults, recovery assumptions, maturity, concentration,
   coverage tests and transaction protections before evaluating a tranche or approving a trade.

Common Mistakes

Using a factor as a one-year default probability. A rating factor has no automatic percentage or
annual interpretation. A probability of default needs a
specified horizon and calibration; a full loss estimate also needs exposure and recovery assumptions.

Mixing agency scales or corporate scorecards. S&P has its own factor definitions. For example,
the July 2, 2024 Carlyle Direct Lending CLO supplemental indenture supplies a separate S&P table and
SPWARF definition. Do not combine those values with Moody’s factors or confuse this portfolio lookup with
weighted business/financial factors used to assess one company’s rating.^[3]

Dropping an unrated loan or treating it as riskless. Absence of a directly usable rating is a
data-resolution problem. The relevant contract may require a derived rating or another specified treatment;
silently assigning zero changes the result without authority.

Calling a lower WARF proof of improvement. A default exclusion, sale or denominator change can
alter the reported average. Reconcile the portfolio movement before attributing it to better borrower quality.

Limitations of WARF

Even identical averages can hide different rating distributions. Retain the underlying mix and stressed exposures,
rather than assuming every asset resembles the average. Agency judgments and contractual mappings also leave
differences within each rating category.

A tranche rating concerns a particular liability and its structural protections. It cannot be read directly from
collateral WARF, and passing a factor test does not guarantee payments. As the SEC explains, credit ratings also
do not measure every market or liquidity risk.^[4]

Sources

1. Barn Owl Funding LLC Credit Agreement, Exhibit 10.1. March 25, 2025;
   Schedule A, Collateral Quality Test; Schedule E-1, rating definitions; Schedule F, factor table, ordinary and
   adjusted WARF, and maximum-factor test. Dated contract example, not current agency-wide policy.
2. Bluejay Funding LLC Credit Agreement, Exhibit 10.1. February 18, 2025;
   Section 1(a), Minimum Diversity Score Test; Schedule H, Diversity Score Calculation.
3. Carlyle Direct Lending CLO 2015-1R LLC, Third Supplemental Indenture.
   July 2, 2024; Appendix A definitions of S&P Rating Factor and S&P Weighted Average Rating Factor;
   Schedule 3, Section 2, S&P Rating Factors.
4. SEC Investor Bulletin: The ABCs of Credit Ratings. October 2013, pages 1–3; relative
   rankings, agency differences, limitations and SEC oversight. Sources accessed November 14, 2026 UTC.