All trades discussed below are paper positions — hypothetical, not executed.
Bank-sector noncurrent loan rates held at 0.98% through March quarter-end, a level that has historically preceded spread widening in private credit as stress migrates from regulated balance sheets into direct lending books. Net charge-offs came in at 0.577% — elevated but not yet recessionary. The regime is one of compression at the edges: not a crisis, but a slow leak.
The most important signal this week isn't a spread; it's a structural one. Our `exit_vs_peer_contradiction` signal — where a BDC exits a position that peer managers continue to hold at par or near-par — is running a 70% hit rate over 60 days on 20 events, generating +3.3% average abnormal return. That is a meaningful divergence-arbitrage signal in a market where price discovery is slow and NAV marks lag reality by at least a quarter. Meanwhile, `v3_borrower_stress` is running a -13.1% average abnormal return on 65 events with only a 46.2% hit rate — suggesting that when our stress model fires, the market is still in denial, and the path forward is almost always lower. Spreads aren't screaming yet, but the borrower-level data is.
This week's list is deliberately thin. The platform returned zero active trade decisions and zero alerts after scoring. That is the system working correctly. We surface five setups from the scorecard history and signal performance — directional reads, not live entries.
BDC UNIVERSE — SHORT BIAS via `v3_borrower_stress`
Sixty-five stress events, 46.2% hit rate, -13.1% average abnormal return over 60 days. When this signal fires on a specific name, the asymmetry is bearish: even when it's "wrong" directionally, the average positive return on the other side is insufficient to offset the drawdown profile. Any BDC with a recent stress flag and material exposure to sponsored software or healthcare services borrowers should be treated as a short candidate until the next filing cycle confirms or denies deterioration. What would prove this wrong: a sustained tightening in SOFR-based floating rates that lifts interest coverage ratios back above 1.5x across the cohort.
EXIT_VS_PEER — LONG DIVERGENCE
The cleanest alpha in our database right now sits in this signal: 70% hit rate, +3.3% abnormal return, 20 events. The logic is simple — when BDC-A exits a credit and BDC-B continues to hold it at par, either BDC-A has information BDC-B lacks, or BDC-B is marking incorrectly. Historically, BDC-A is right 70% of the time. The trade is to be short the holding BDC's NAV-sensitive equity, or long the exiting BDC on the thesis that their credit discipline is superior and will show up in future book value stability. The falsifying condition: BDC-B's borrower cures, refinances, or is acquired, validating their mark.
PLATFORM BOOK — NEUTRAL/MONITOR
Our internal book value sits at $100,293 as of July 17, 2026 — essentially flat to a round number, which is an uninformative mark in isolation. Without NIM lead-lag data populating this week (the `nim_leadlag` array returned empty), we cannot confirm whether net interest margin is expanding or compressing relative to historical seasonal patterns. We are not adding directional exposure here. Watch the next NAV print.
DEBT-TO-EQUITY CONVERSION — AVOID
This signal has a 28.6% hit rate and -2.3% average abnormal return on 7 events. Debt-to-equity conversions inside BDC portfolios are almost universally distress outcomes dressed up in restructuring language. The market often treats conversion announcements as "proactive management" — we treat them as confirmatory deterioration. No current active position here, but any BDC announcing a material conversion in the next cycle should see immediate scorecard demotion.
EXIT_AT_LOSS — NEUTRAL
One hundred sixty-eight events, 49.4% hit rate, +0.18% average abnormal return. This signal is essentially a coin flip with noise. Exits at a loss tell you a BDC is realizing discipline, but the forward price action is random at the 60-day horizon. We are not building a strategy around it.
The platform passed on everything actionable this week — no decisions, no alerts. That is not a failure; it is the scorecard doing its job when the data does not support conviction. We did not force a trade because the calendar said to write one.
The next meaningful read will come from Q2 10-Q filings hitting EDGAR through mid-September — watch borrower-level PIK income as a percent of total investment income for early stress migration.
Screening research, not investment advice. All data from public SEC filings.