A credit spread is the extra yield investors demand to hold a corporate bond instead of a Treasury. Default is in that extra yield. So is the chance you get stuck in a bond nobody wants, and the chance the company calls the bond back on the day that suits the company. Equities often move on the same afternoon. Shareholders only get what is left after those lenders are paid. The interest bill the company actually writes stays where it is until a coupon resets or a new bond is sold.
I watch two series. Both are from ICE Data Indices, and FRED publishes both daily. The investment grade one is the ICE BofA US Corporate Index option-adjusted spread, series BAMLC0A0CM. The high yield one is the ICE BofA US High Yield Index option-adjusted spread, series BAMLH0A0HYM2. FRED quotes them in percent. Traders still shout basis points, and I have misread this page because of that. A print of 3.44 means 3.44 percentage points over Treasuries, which is 344 basis points. One percentage point is 100 basis points. Mix the units up and the market looks either on fire or asleep.
What gets averaged
Option-adjusted means the calculation takes out the value of embedded options. Mostly that is the issuer's right to call the bond and refinance it. Subtract a Treasury yield from the raw bond yield and the call is still in the number you have left. The OAS is an attempt to peel the call off and leave the credit and liquidity piece. A model sits in the middle of that attempt. You cannot look at the coupon and read the OAS off it.
The investment grade index holds US dollar corporate bonds rated BBB or better on an average of Moody's, S&P and Fitch, with more than a year left, a fixed coupon, and at least $250 million outstanding. The high yield index holds bonds rated BB or below, and the minimum there is $100 million. Both want an investment grade country of risk. Both skip defaulted bonds. Both are weighted by the amount outstanding, so a big issuer moves the print more than a small one. A single name can be in real trouble while the index looks fine. I see this mistake more than any other. Someone treats BAMLH0A0HYM2 as "the junk bond market," and the next sentence is about their company.
The series are daily closes, and they are not seasonally adjusted. Since April 2026, FRED has distributed only about three years of each series and tells you to go to ICE for the rest. The file I can cite runs from November 3, 2023 through the October 30, 2026 close. Look for the 2008 peak in that download and it isn't there. March 2020 isn't either. I am not going to type a number for either one from memory. I would be making it up, and I have wanted to. What I do remember, and still use, is the shape. High yield moved more than investment grade. Share prices were already falling. Later, after policy backstops were in place, the spread came back in. I leave the cell blank when the download has no peak to put there.
Two scares that round-tripped
Inside the window FRED still publishes, the violent days are easy to find. They are also smaller than the folklore. On January 22, 2025, high yield OAS closed at 2.59 percent, the low of the file, and investment grade closed at 0.81 percent. On April 7, 2025, high yield closed at 4.61 percent, the high of the file. From the January low that is a widening of 2.02 percentage points, or 202 basis points. Investment grade on April 7 was 1.20 percent, and it ticked to 1.21 percent on April 9. From the January low, investment grade widened 0.40 percentage points at the most. High yield did the panicking. Investment grade widened, then barely moved.
Most of the high yield move was a few sessions. Not a slow grind. The April 2, 2025 close was 3.42 percent. April 3 was 4.01. April 4 was 4.45. April 7 was 4.61. That week lined up with the broad US tariff announcement on April 2. I keep that as context. I can't show that tariffs were the only seller. By June 30, 2025, high yield was back to 2.96 percent and investment grade was 0.86. The market charged a scare premium and then gave most of it back inside a quarter.
August 2024 was the same pattern at a smaller size. I like it as a drill, because once it was over nothing about it felt historic. On July 31, 2024, high yield was 3.25 percent and investment grade was 0.97. On August 5 they were 3.93 and 1.12. On August 16 they were 3.29 and 0.98. If your rule is "sell the equity book when high yield jumps half a point," you sold a move that was gone in eleven days. I don't think that rule survives contact with this file.
| Close | High yield OAS | Investment grade OAS |
|---|---|---|
| Nov 3, 2023 | 4.04 | 1.29 |
| Aug 5, 2024 | 3.93 | 1.12 |
| Jan 22, 2025 | 2.59 | 0.81 |
| Apr 2, 2025 | 3.42 | 0.96 |
| Apr 7, 2025 | 4.61 | 1.20 |
| Apr 9, 2025 | 4.37 | 1.21 |
| Jun 30, 2025 | 2.96 | 0.86 |
| Jan 22, 2026 | 2.64 | 0.73 |
| Oct 30, 2026 | 3.44 | 0.94 |
Figures are percent, daily closes, from FRED series BAMLH0A0HYM2 and BAMLC0A0CM. November 3, 2023 is the first row of the current file, and it also happens to be the widest investment grade print in it, at 1.29 percent. High yield that day was 4.04, which is wide for this window and still short of the April 7, 2025 high. The two series peaked on different days. They bottomed on different days too. High yield's low is January 22, 2025, at 2.59. Investment grade's low is January 22, 2026, at 0.73, when high yield was already back up at 2.64. Follow only the junk series and you miss a year in which the higher-quality market kept getting more comfortable.
The latest close in the file, October 30, 2026, was 3.44 percent high yield and 0.94 percent investment grade. High yield sits 0.85 percentage points above its low. Investment grade is 0.21 above its low, and 0.35 under the widest mark in the file. Against January 2026 the market is charging a bit more. 4.61 felt sharp because January had been 2.59. I can't tell from the level alone whether the new-issue market shut.
The level, the change, and your own bond
I line the level up against the last few months. FRED won't give me a 2008 number right now, so I leave 2008 out. Then both series, same day. If high yield is the one that widened, weaker borrowers are what I'm arguing about. April 2025 was the other picture. Investment grade widened too, which puts risk appetite and liquidity in the story, and the investment grade holdings with them. Even so, the pain sat mostly in high yield. The bond in front of you may be missing from the index. It may be a small weight. It may be moving because of a lawsuit, a merger, or a downgrade. An average will sit on top of that and hide it.
Coupon is a different number. I confuse the two when I go too fast. Take a new fixed-rate bond. Yield equals a Treasury yield plus a credit spread, before fees, and before whatever concession the syndicate needs that morning. Treasury yields can fall on a day when spreads rise. The coupon the treasurer actually prints can still come out close to the old one. BAMLH0A0HYM2 won't show you that Treasury half. Policy rates, and the lag before they reach cash interest, are on another clock. More on that in what a central bank rate decision means for your holdings and when a rate move reaches the cash.
The share price, then the interest bill
Same afternoon, look at the share price. Equity is whatever is left after the lenders. Lenders mark the debt riskier, and the shares usually get marked down with it. That can happen before any borrower has missed a payment. Shares can also snap back as fast as the spread did. August 2024. April 2025. I haven't rebuilt a portfolio off a three-day widening. I do write down which positions need the bond market to stay open.
Refinancing takes longer. Cash interest expense changes when floating debt resets, when commercial paper rolls, or when a maturity is paid off with a new issue. Until that date a wider OAS sits in a footnote, and the income statement still shows the old line. Oracle is the case I keep for this, because the buildout is funded in the market and the old coupons don't care what high yield did last Tuesday. What the spread can change is the next deal. I use the maturity table for the date that deal has to happen.
Banks meet the spread from the other side, and the sign is messy. JPMorgan Chase's FY2025 Form 10-K, year ended December 31, 2025, says that when credit spreads widen it becomes more expensive for the firm to borrow, and that the firm's own spreads move both for reasons specific to the firm and for general economic and geopolitical reasons. It also says changes in those spreads can hurt earnings on certain liabilities carried at fair value, including derivatives. I'm reading the bank's own borrowing spread there, and the bank's own marks. BAMLH0A0HYM2 doesn't fall out of that paragraph as an earnings-per-share number. See JPMorgan.
The loan book is a third channel. It doesn't line up with the index either. I checked, because I assumed it would. The same 10-K reports a provision for credit losses of $14.2 billion in 2025, net charge-offs of $9.8 billion, and an allowance for credit losses of $31.2 billion at year end. The consumer piece of the provision was $11.5 billion, and the firm says $2.2 billion of the consumer allowance build was related to the Apple Card transaction, alongside card growth and the macroeconomic outlook it uses. High yield spent a lot of 2025 near 3 percent, spiked to 4.61 for a moment, and was under 3 by the end of June. The provision was still $14.2 billion. A tight index and a $14.2 billion provision sat in the same year, and the April spike doesn't account for an allowance that includes a portfolio deal. If you own the bank, read the allowance footnote. I have misread a bank by stopping at the ICE average.
One place in the 10-K does price a 100 basis point spread move, and it is easy to grab for the wrong series. In the mortgage servicing rights note, the weighted-average option-adjusted spread assumption at December 31, 2025 was 6.14 percent. The firm says a 100 basis point adverse change in that assumption would cut fair value by $394 million, and a 200 basis point adverse change would cut it by $757 million. That OAS is an input to a servicing model. BAMLC0A0CM is a different series, and so is the high yield index. A bad day in junk bonds doesn't move this line by $394 million. I keep the figure because it shows how a bank discloses spread sensitivity when it actually has one, with an instrument, an assumption, and a dollar result. Most equity writeups skip all three.
Bank of America is the same kind of problem with different numbers. The rate tables in its filing tell you how net interest income behaves if the yield curve jumps in parallel. They have nothing to say about card charge-offs if high yield stays at 4.61 for a year instead of a week. A headline that only says "rates" will staple the two together. I keep a separate note for each, and only then do I decide whether the holding is "sensitive."
Borrowers who aren't banks pass the spread through to revenue, and they do it late. The customer might be borrowing for a data center. Or a house. Or a fleet. If the new loan is expensive, or if the loan simply isn't there, the order can wait. That shows up as the seller's sales. It doesn't show up as the seller's interest expense. I miss it when I only open the debt footnote of the company I own. Sector tendencies are a starting map. They won't tell me how this customer funds the order. That check is separate. The map is in which sectors react to which shocks.
What I write next to the position
- One word next to the position. Borrows, lends, or sells to someone who borrows. A wider spread hits those three on different dates.
- BAMLH0A0HYM2 and BAMLC0A0CM, same day. Write down which one moved. Look again a month later. The April 2025 high yield spike was mostly gone by June 30.
- FRED is in percent. From 2.59 to 4.61 is 2.02 percentage points. Call it 2 basis points and the units are swapped.
- Own a bank? Open the provision, and the sentence about the firm's own credit spread. JPMorgan's FY2025 provision was $14.2 billion, in a year when the high yield index round-tripped. The allowance comes from the filing.
- If the company has to issue, the spread is the backdrop for the next coupon. The date is the maturity table. Oracle is the page I use. Old fixed coupons stay put when the index moves.
- Skip the 2008 peak until you have it from ICE, or from a FRED vintage that still contains it. The public file since April 2026 doesn't have it.
I decide on a quiet day what would count as early, and what would count as late. Before that, I don't treat the spread as a warning. Three days up and back is nerves. Still wide when the next bond has to be sold, and I'm looking at cash. One headline number gets used for both. A holding that refinances next month is having a different week from one that only watched the spread move.
Educational reading of credit spreads and how they reach listed companies. Not investment advice. OAS levels are ICE Data Indices series as published on FRED (BAMLH0A0HYM2 and BAMLC0A0CM), daily closes from November 3, 2023 through October 30, 2026. Company figures are from JPMorgan Chase's Form 10-K for the year ended December 31, 2025. Spreads, allowances and maturity schedules change.