Andy Wells

Remember the October 2025 CPI print that never happened? PPI doesn't have that problem. Producer prices still got collected after the shutdown, just late. So Friday's PPI has an official year-over-year number. Consensus is 6.6%, which would be the highest since Nov 2022. In September, PPI ran 2.60 pts hotter than CPI, the widest gap since Apr 2022. That gap closes through margins or through CPI. After three straight 0.3% core CPIs, I'd bet on CPI. With a December hike now looking assured off CPI, will a hot PPI tip it from 25bps to 50? #PPI #Inflation

Is retail margin a problem? $1.48T of margin debt vs $434B of cash at the brokers (FINRA, Sept). That's $3.41 borrowed per $1 of cash. At the 2021 peak it was $2.19. The midterm honeymoon lasted one day (let's hope that changes). The order a margin call travels: chips, then anything liquid, then credit. Today is step two, with all 11 sectors red. Credit is the last stop. Most of the widening so far has occurred in specific names like CoreWeave. Want to know if this goes structural? Watch the high yield spread. HY OAS is 340, exactly its 2020s median. If that breaks 381 (the 2020s average), this stops being a tech problem. #MarginDebt #Credit

Stocks are on pace for their worst day since April 2025 and bonds finally acted like bonds. 2Y -11 bps. 10Y -6bps. 30Y -3bps. Oil was up 2.5% and the 2-year still ripped. The market is shaving the December hike. The long end is following the risk-off sentiment. Payrolls tomorrow!

Three things moving bonds this week. Which day matters most? Likely not Tuesday. 1/ Midterms (Tue): Kalshi D sweep 71%, House ~93%. A GOP Senate (27%) is the only real surprise left. 2/ Refunding (Wed 8:30 ET): dealers expect softer "next several quarters" coupon guidance. Citi expects 20s/30s cuts starting this month. 3/ Payrolls (Fri): 55K consensus, 4.2% UE. Dec hike ~85% priced, so a soft print could rally 2s. 30s rallied 30 bps in October to 5.31%. I expect softer guidance and no size cuts yet.

62% of Bloomberg Markets Pulse respondents say stricter AI regulation under a new Congress hurts stocks. Kalshi: House flip ~95%, Senate ~70%. No veto-proof majority, so AI law likely waits for 2029. What gets pulled forward is oversight, and preemption likely stalls, so state AI laws stand. That's a cost-of-capital story. 78 of 91 hyperscaler bonds issued in 2026 trade wider than issue. #AI #Midterms #Credit

What happens to stocks when markets never close? ICE is bringing 24/7 perps onshore, and there's already a preview offshore: stocks and stock indexes are 85.7% of non-crypto perp volume on Hyperliquid. Proponents talk of more liquidity and better price discovery. Early read on what we could expect: Monday opens get priced over the weekend. After a Saturday war headline in February, CME crude opened Sunday night right at the oil perp's price. Leverage migrates. $200B in leveraged ETFs and 0DTE at 36% of options volume now compete with a product that never resets and never expires. Weekends get jumpier. Chip-stock perps lost $51.6M to liquidations in 10 minutes on Saturday, Oct 17. What will this displace? Likely 2x levered single-stock ETFs for a start.

Fed held 10-2. December hike odds fell 90.7% to 85.3%. 2s rallied 1.2 bps. 30s sold off 5.3. The path came in and the long end went up anyway. That’s basically all term premium. The Kim-Wright vs ACM term premium gap was ~25 bps going in, near a 3-year high. My guess is today made it wider.

1ET 5-Year Auction Report The 5-year cleared at 4.845%, 18.8 bps below September. Tail was 0.4 bp. Here is the important part (who bought it): Dealers took only 10.6%. Directs stepped up to 29.4%. Indirects were light at 60.0%. So domestic real money carried it. Let’s rate it a 6/10. Treasury got its money without paying up. Meanwhile, Anthropic’s IPO filing shows a $63 billion Broadcom chip lease, financed by banks and Blackstone, with Broadcom backstopping the collateral. The money’s there. But AI platforms, chipmakers, banks and private credit are getting awfully tangled, and that deserves a bond investor’s attention. Where I sit: overweight 3 to 7-year Treasuries, constructive on intermediate TIPS, cautious on 20 to 30-year duration. Next up is tomorrow’s FOMC presser.

Why would Meta sell $95B of stock when it can borrow at investment grade? Because the bond market started charging more. Hyperscaler bond issuance went from ~$35B a year (2020-24) to $93B in 2025 and $132B through July 2026. Spreads widened ~30 bps YTD vs 2 for broad IG, and order books slid from ~5x covered in Feb to under 2x by July. Now stack up what's on the table: Meta $95B, Anthropic's IPO (reported up to $100B), SpaceX's $40B GPU lease, a reported $30B OpenAI round. Roughly $265B, of which $35B has actually launched. For bondholders this is the good outcome. Equity sits under the bonds and takes supply off the IG calendar. The valuation pressure lands on shareholders instead.

The 10-year rallied 16 bps today to 4.96%. Did it just get rich? A little. Real growth (2.2%) plus core PCE (3.0%) puts fair value near 5.20%. That's 24 bps under, on a day with no major data and Brent at $112. The drop looks like it came out of the real-yield leg. AI stocks sold off, and the cost of hedging a bigger bond rally hit its highest since March. A fear bid with shorts covering on top. Oct 29 brings Q3 GDP and core PCE. Look to those for guidance on whether this rally can go deeper into the 4s. #Treasuries #Rates

1ET 5-Year Auction Report The new 5-year TIPS cleared at a 2.367% real yield. Highest at auction for this term in three years. It tailed the when-issued by 0.7 bps. A little sloppy. Here’s the context: Coming into the week this looked like a 2.5% print, maybe even a 20-year high. Treasury’s own curve estimate had the 5-year real yield at 2.55% at yesterday’s close. Then today went risk-off and everyone piled into Treasurys. So real money had a limit order sitting at 2.50% and watched the market gap right past it. (Super annoying I guess) So who bought it? Indirects came in below average. Covers were light at 2.41, the lowest since April 2025. Dealers only got stuck with 5.5% though, well under their recent average, and directs picked up a good chunk of the slack. Price sensitive buyers decided not to be bagholders at 18bps below yesterday’s real yield. That reads like domestic buyers still showed up, just at a worse level than they’d been hoping for. Breakeven came out around 2.41% with the 5-year note at 4.78%. Overall I’ll rate it a 5.5/10. Soft demand. Tail. But Treasury got it’s $26 billion done.

Huge rallying happening in treasuries. Today’s move is investors buying protection against the AI rally being overdone. Is this a disinflationary rally? No. Is there signs of a recession? No. The rally is predominantly in real yields. From October 5 through October 20: Nominal 10-year yield: −19 bp Real 10-year yield: −18 bp Inflation breakeven: −1 bp That makes approximately 95% of the decline attributable to real yields. Probably some more short covering still on treasuries. How long will today’s rally run? However long the Nasdaq selloff lasts. Market is turning risk-off.

Taxable bond ETF issuance went from $20.8B to $11.0B in one week. Muni ETFs went from $4.6B to $8.03B. (ICI, week ended Oct 14) Munis were 18% of bond ETF issuance in mid-Sept. Last week, 42%. Some of that is tax-loss swapping. I doubt all of it is. The AI build keeps leaning on the taxable market: $80B of data center bonds this year, and $37B of it is junk. I'd rather be in 3 to 7-year Treasuries than 20s and 30s right now, and pickier in credit.

How much of a 7.40% mortgage is the 10-year? About 71%. Build it from the bottom (week of Oct 9): 10-year: 5.24 MBS spread: 1.12 Fannie/Freddie g-fee: 0.66 servicing: 0.25 lender margin: ~0.13 = 7.40% Notice what isn't there. No Fed rate cuts or hikes. Only two pieces trade freely, the 10-year and the mortgage spread, and they're 86% of the rate. Want to know where mortgages are headed? Watch the 10-year, inflation and the MOVE index.

Who does a bond index own the most of? Whoever borrows the most. There's no expert sitting on top deciding if that's a good idea. It's a pass-through. AI-related bonds are about 4% of the US high yield index. But they were 40% of 2026 net new HY issuance (Guggenheim). Each deal rolls in at month-end and passive money buys its share. Meanwhile the paper is repricing against us. HY OAS 2.60% on Aug 28, 3.31% Friday, widest since March 30. CoreWeave 5Y CDS past 1,100 bps. Chips bounced back Monday. Credit takes longer.