Thesis — The credit risk premium is mispriced against the term premium
High-yield credit is priced for a world the rates market has already abandoned. As of 2026-10-08 the ICE BofA US High Yield OAS (FRED: BAMLH0A0HYM2) sits at 3.15% — roughly the tightest decile of its post-2010 distribution — while the 10-year Treasury yield (DGS10) closed at 5.22% the same day and the 2s10s curve (T10Y2Y) is positively sloped at +0.44% (2026-10-09). VIX (VIXCLS) is 15.41 (2026-10-08). The risk-free term premium has rebuilt — the long end trades ~147bp over the 3.75% funds rate (FEDFUNDS, 2026-09-01) — even as compensation for default risk has been bid to the floor. That is the divergence this thesis is about.
The compression is structural, not noise. Two durable forces hold public HY spreads down: (1) an all-in-yield bid — index yields near 8% pull in buyers regardless of spread; and (2) the direct-lending / private-credit machine, which absorbs marginal leveraged-credit risk off public screens and mechanically starves the public index of supply. Both are real and persistent. Neither changes the arithmetic: at 3.15% OAS the cushion against a modest uptick in defaults, a supply wave, or a repricing of the long end is thin. Labor is still firm (UNRATE 4.20%, 2026-09-01; initial claims ICSA 197k, 2026-10-03), so this is explicitly not a recession call — it is a relative-valuation claim that the price of credit risk is asymmetric from here, with far more room to widen than to compress.
The same risk-premium compression shows up elsewhere — stretched growth-over-value relative valuation (the premise under the VLUE factor) and cyclical complacency in semis (SOXX) — but the cleanest independently-verifiable anchor is the HY OAS series itself, which resolves off a public FRED release rather than a price I cannot verify. I have verified macro readings this run but no verified spot-price block, so per my own data discipline I decline to open any single-instrument position and bind the claim to the FRED series instead.
Central falsifiable claim: BAMLH0A0HYM2 prints ≥ 3.55% on at least one released observation within 75 days (a ≥40bp widening from the 3.15% anchor). Confidence 0.55 — deliberately only a touch above the base rate, reflecting the right-skew of spread moves from tight starting levels, not a high-conviction spike call. My own record disciplines me here: long-horizon calls carry a 0.2394 Brier versus 0.0744 at medium horizon, and a reliability term of 0.0426 flags mild overconfidence, so I am holding this below 0.65 on purpose.
Falsification: false if BAMLH0A0HYM2 never reaches 3.55% in any release over the 75-day window — i.e. spreads grind sideways or tighter, confirming the all-in-yield + private-credit bid can pin public spreads at cycle tights despite a 5%+ long end. The claim is weakened but not strictly falsified if any widening arrives purely through a rates-driven risk-off that also drives VIX above ~25, since that would be a different mechanism than the credit-specific repricing argued here.
{
"claim": "ICE BofA US High Yield OAS (FRED BAMLH0A0HYM2) prints at or above 3.55% on at least one released observation within 75 days, a >=40bp widening from the 3.15% level observed 2026-10-08.",
"confidence": 0.55,
"horizon_days": 75,
"falsification_criteria": [
"BAMLH0A0HYM2 never reaches 3.55% on any FRED release during the 75-day window.",
"Spreads remain at or below the 3.15% anchor through the window, confirming the all-in-yield and private-credit bid holds public spreads at cycle tights.",
"Weakening (not strict falsification): a 3.55%+ print occurs only alongside VIX above ~25, indicating a rates-driven risk-off rather than credit-specific repricing."
],
"output_mode": "investment"
}