Methodology

The Evergreen Private Markets Index family provides transparent, rules-based benchmarks for private credit and equity markets as accessed through registered evergreen vehicles. Unlike traditional private markets indices that rely on voluntary manager submissions, these indices are constructed entirely from mandatory SEC filings, ensuring complete and unbiased coverage of the registered vehicle universe.

Four indices are published, each tracking a distinct segment of the private markets:

• Direct Lending Index -- The largest position-level benchmark for middle-market direct lending
• Direct Equity Index -- Direct equity co-investments and minority stakes in private companies, held by registered closed-end vehicles
• Private Credit Fund Index -- Allocations to private credit strategies including CLOs, direct lending funds, and specialty finance platforms
• Private Equity Fund Index -- Allocations to private equity strategies including buyout, growth equity, and venture capital funds

All data is sourced from SEC EDGAR, the public repository of regulatory filings. Three filing types provide the raw data:

BDC XBRL (10-K/10-Q)

Business Development Companies file annual (10-K) and quarterly (10-Q) reports that include a Schedule of Investments in structured XBRL format. The SEC began requiring inline XBRL tagging for BDC investment schedules starting in 2022-2023. Each position is tagged with fair value, cost, interest rate, maturity, and other attributes using a typed dimension (investmentIdentifierAxis).

N-PORT (Quarterly TSV)

Registered investment companies (including interval funds and tender offer funds) file monthly portfolio holdings on Form N-PORT. The SEC publishes quarterly bulk data sets containing all N-PORT filings in tab-separated format. Each holding includes CUSIP, issuer type, asset category, fair value, and additional fields for debt securities.

N-CEN and N-2

Form N-CEN provides census data for registered investment companies, including fund type classification. Form N-2 cover pages contain checkboxes identifying whether a fund is a BDC, interval fund, or closed-end fund. These are used for universe construction, not holdings extraction.

The investment universe is built using six independent discovery methods, then cross-validated against third-party lists:

1. N-54A/N-54C elections -- BDCs that have elected or withdrawn BDC status with the SEC
2. 814- file numbers -- CIKs with Investment Company Act file numbers in the 814 series
3. SEC BDC data set -- CIKs appearing in the SEC's structured BDC data sets
4. N-CEN classification -- Funds tagged as interval funds or N-2 registrants in N-CEN filings
5. N-2 checkbox parsing -- Automated parsing of N-2 cover page checkboxes to identify interval funds, BDCs, and closed-end funds
6. EFTS text search -- Full-text search of SEC filings for “tender offer fund” and related terms

Holdings are extracted from two sources and unified into a single dataset:

BDC Holdings

XBRL instance documents are downloaded for each BDC's 10-K and 10-Q filings. The parser extracts facts tagged under the investmentIdentifierAxis dimension, mapping XBRL concepts to standardized fields (fair_value, cost, interest_rate, principal_amount, etc.). A concept mapping table handles variations in XBRL taxonomies across filers.

N-PORT Holdings

Quarterly TSV data sets are downloaded from the SEC DERA website. Holdings are filtered to universe CIKs and mapped to the unified schema. Balance and unit fields are converted: PA (par amount) maps to principal_amount, NS (number of shares) maps to shares_held.

Unification

BDC and N-PORT holdings are combined with consistent field names, rate harmonization (BDC rates converted from decimal to percentage), and cross-source deduplication. When a BDC also files N-PORT, the BDC XBRL data is preferred for richer field coverage.

Each holding is classified into one of four index categories based on asset category and issuer category:

Additional heuristics reduce the unclassified rate: BDC financial field fallback (positions with interest rates default to LOAN), named co-investment reclassification (fund positions with identifiable operating company names reclassified to equity), and expanded keyword lists for credit/PE fund detection.

To compute returns, the same position must be linked across consecutive quarters. A four-tier matching cascade is used:

1. Tier A: Within-filing comparatives -- BDC XBRL filings contain both current and prior-period facts under the same investmentIdentifierAxis value. These are filer-matched pairs requiring no external matching.
2. Tier B1: CUSIP matching -- N-PORT holdings with CUSIP identifiers are matched across quarters on the same CUSIP within the same CIK.
3. Tier B2: Exact name matching -- Positions are matched by exact issuer_name within the same CIK across adjacent quarters.
4. Tier C/D: Normalized and fuzzy matching -- Fallback matching using name normalization and Jaro-Winkler similarity with fair value proximity guards.

Total return for each position is decomposed into three components, following the convention of public credit indices:

Capital Return (Price)

Per-unit price return isolates price changes from quantity changes. For loans: price = fair_value / principal_amount. For equity: price = fair_value / shares_held, with fallback to fair_value / cost. This prevents amortizing loans from being penalized and new purchases from inflating returns.

Income Return

Estimated coupon accrual based on a three-tier rate imputation: (1) direct interest_rate from the filing, (2) basis_spread plus implied SOFR from peer filers, (3) same-filer median rate. Income is accrued proportionally to the holding period.

Index Aggregation

Index-level returns are computed as fair-value-weighted averages across all constituents. Equal-weighted returns are also published. Index levels are chain-linked from a base of 100. A minimum of 10 constituents per quarter is required for a valid index observation.

The indices rebalance quarterly, aligned with SEC filing deadlines. BDC 10-K/10-Q filings have a 60-day deadline after fiscal period end; N-PORT filings have a 60-day deadline. Data is available with a one-quarter lag.

New positions entering the index (no prior-period match) contribute to AUM statistics but not to return calculations for their first quarter. Exiting positions (matched in the prior period but absent in the current) are included in returns for their final period.

• Coverage starts ~2022 -- BDC XBRL tagging was phased in by the SEC starting in 2022. Pre-2022 filings are unstructured HTML.
• Quarterly frequency -- Unlike daily public market indices, these indices update quarterly due to the filing cadence.
• Fair value estimation -- Reported fair values reflect fund-level mark-to-model estimates, not market transaction prices.
• Survivorship bias -- Funds that de-register or stop filing exit the index. Historical returns include only active filers.
• Income estimation -- Coupon income is estimated from filed rate data, not actual cash received. PIK and amendment effects may not be fully captured.
• Incomplete field coverage -- Portfolio characteristics such as weighted average coupon, spread, and maturity are based on positions where the filer disclosed the relevant data. Not all filers tag all fields, so reported metrics may reflect a subset of the index universe.
• No leverage adjustment -- Returns reflect gross portfolio performance, not fund-level returns which may include leverage effects.