How it actually works
A bond yields more than a Treasury of the same maturity because you might not get paid back. That gap is the spread, and it is the whole reason anyone lends to a company rather than to the government.
The complication is that many bonds contain options. The issuer can often call the bond early and refinance if rates fall. Mortgage borrowers do the same thing by prepaying. Occasionally the investor gets a put. Each of those features is worth money to somebody, and the price of the bond already reflects it, so the crude yield gap over Treasuries mixes two different things together: payment for credit risk, and payment for the option you sold without quite noticing.
separates them. A model generates a large number of possible future interest-rate paths, values the bond's cash flows along each path while assuming the issuer exercises its option whenever that is the sensible thing to do, and then solves for the single constant spread that, added to the whole benchmark curve, makes the model's price match the actual market price. That constant is the OAS. It is the compensation left over once the option has been paid for.
An illustration, with invented round numbers. Say a callable corporate bond yields 5.60% while the matching Treasury yields 4.60%. The nominal spread is 100 basis points. If the model reckons the call feature the investor has effectively given away is worth 25 basis points a year, the option-adjusted spread is 75 basis points. That 75 is the number you can honestly compare with a bond that has no call at all. For a plain non-callable bullet bond there is nothing to adjust for, so OAS collapses to the ordinary spread over the curve.
Two things follow. First, OAS is a model output, not an observation. It depends heavily on the assumed volatility of interest rates, and reasonable people using different assumptions get different answers for the same bond; "option-adjusted" means "adjusted according to somebody's view of the future". Second, index OAS — the aggregate figure quoted for the investment-grade or high-yield market — is a size-weighted average across thousands of bonds. It tells you about the typical bond in the index, not the bond being sold this morning.
Why it matters to this crash
Index OAS is the benchmark everything else gets measured against, which makes it the thing to look at when a rating and a price disagree.
On 19 August we wrote about Project Odyssey, the roughly $3.9bn five-year deal from Blackstone-backed QTS funding a Microsoft-linked data centre. Initial price talk was a yield of about 7.63%, on paper expected to be rated Baa3 by Moody's and BBB− by Fitch. Against two-year Treasuries at 4.19% and tens at 4.72%, that is on the order of 300 basis points of spread. The investment-grade index option-adjusted spread at the time was 81 basis points. Roughly four times the average, with an investment-grade rating attached. That comparison is our arithmetic, not the bank's, and it is a crude one: a single new issue's yield-minus-Treasury gap is not measured the same way as a model-derived index average, and and call features differ. The gap is still far too wide for the methodology to explain away.
That is the use of the number. On its own, 81 basis points says credit conditions are calm. Placed next to a deal clearing at 300, it says something more specific: the calm is in the average, and the average is made of bonds that were priced a while ago. The marginal deal — the one that has to find real buyers today, in — is where the disagreement shows up first. Tight index and junk pricing on new paper are not contradictory readings. They are an early and a late reading of the same thing.
What would make this dangerous
The investment-grade index OAS moving materially off 81 basis points, its August 2026 level. A doubling would mean the average has caught up with the margin rather than the other way round, and every issuer with a refinancing to do in the next two years would be doing it at a very different price.
More data-centre and AI-infrastructure paper clearing at multiples of the index OAS while keeping its investment-grade ratings. One deal is a negotiation. A queue of them is the ratings framework failing in public, and it would mean index-tracking investment-grade funds are holding a category of risk their mandate did not contemplate.
The index OAS staying tight while dealer bid-offer widens and secondary trading thins out. Index OAS is derived from prices, and prices on bonds nobody is trading are estimates. A tight spread on stale is not the same thing as a tight spread, and the difference only becomes visible when someone needs to sell.