For every pair of BDCs, what fraction of portfolio is the same borrowers? And for every borrower held by multiple BDCs, how widely do those BDCs disagree on what the credit is worth? This module answers both questions in a single analytical surface. The cross platform overlaps reveal systemic concentration risk. The mark dispersion finds the credits where someone's valuation is wrong.
Per the Raymond James and PIMCO convention, pairwise overlap between two BDCs is the sum of min(weight_A, weight_B) across every shared borrower, where weight is the holding's fair value as a percent of the BDC's portfolio. The result is a number from 0% (no shared borrowers) to 100% (identical portfolios).
Borrower names are normalized across roughly 25,000 raw labels using lowercase, punctuation stripping, and trailing corporate suffix removal (Inc, LLC, Corp, Holdings, LP, etc). About 24,700 unique normalized entities result. Schedule of Investment section headings ("Non-Controlled Non-Affiliated Portfolio Investments") are filtered out as they're regulatory grouping labels, not real borrowers.
Equity holdings (warrants, preferred stock, common stock) are excluded from the overlap computation per Raymond James convention, because equity valuations behave differently from debt valuations. Debt instruments only · 1L, 2L, mezzanine, unsecured.
When two BDCs hold the same debt of the same borrower and mark it at materially different values, one of them is wrong. The most likely error is at the lender that took its position via participation or syndication, since the lead arranger has better real time information on borrower performance.
The famous case is CION marking Isagenix at 98 while peers marked at 86 to 88 · a marker spread of roughly 10 points that CreditSights flagged six months before Isagenix went into restructuring. Once the company defaulted, the marks converged downward to roughly 60.
Marks outside the 30 to 130 range are excluded as data quality outliers. The dispersion metric is the population standard deviation across all BDCs reporting a mark for that borrower. We use the largest fair value position when the same BDC reports multiple holdings for the same borrower (e.g. a tranche split).