Harley Bassman — In FED We Trust

01:39 Should have been 50, one and done

• His last two commentaries argued for a 50bp hike. 25, 50 or even 100 won't change the economy or inflation; what matters is regaining the market's trust that "there's an adult in the room."
• "It should have been zero or 50." A 25bp hike with another promised means 12 more weeks of torture (no move in October, before the election).

02:26 Warsh is weaning markets off forward guidance

• Warsh is trying to move away from forward guidance and dot plots, back to markets assessing their own risk. Coming off guidance is "like coming off smoking… heroin"; volatility rises on the way, and cold turkey would have been better.

04:42 The Supreme Court carved out the Fed

• The Court ruled the president can hire and fire across the executive branch, but carved out the Fed as "too special": the governor on the world reserve currency. That raises the stakes on the Fed's credibility while trust elsewhere in government is low.

06:01 Fed–Treasury: symbiosis is good enough

• He doesn't want a public 1951-style accord. If Bessent and the Fed work symbiotically and show "a path to a good result," investors and foreign governments will see the adults managing the reserve currency.
• Warsh swallowed hard and did 25 to get a 12-0 vote. "Having unanimity was a good result."

08:21 Why trust is damaged: profits, sentiment, housing

• He can't make an economic case to hike, and could argue for holding flat. Corporate profits as a share of GDP doubled from 6% to 12%; the S&P and Michigan sentiment have opened an "alligator jaw," a measure of distrust.
• Housing affordability is the worst in 30 years; the homeownership rate should be 61–64%; the median first-time buyer's age jumped from 32–33 to 38 in five or six years.

10:24 Non-bank lending blunts the Fed

• Private credit "probably" becomes a problem someday. The bigger point today is that non-banks disintermediate the banks that borrow from and price off the Fed, so the policy rate transmits less and trust matters more.

11:21 Hyperscalers: price-insensitive borrowers

• Slide 9 projects hyperscaler borrowing of "750 million" against Treasury's "2 billion" (as captioned; units are probably billions/trillions), about 30–35% competition for money. They'll borrow at 6, 8, 10 or 12% in an "existential race." The Fed can't stop it, and it helps push rates up.
• It echoes the dot-com era: ten companies in one niche, one wins, and you can't know which (Pets.com didn't).

13:24 The AI slow-down feud is political spectacle

• On Amodei's call to slow AI versus Trump and Sacks's "full steam ahead": "I think the latter" (it will pass). He expects the AI labs to adopt internal controls regardless of the politics, and to find a middle ground.

14:52 Hyperscaler bonds are money good; the equity is another matter

• They have "underlying cash machines," so higher rates are possible but not bankruptcy. The debt is fine even with another 50–200bp of spread. Profits may fall from the spending: "the equity I can't comment on."
• Anthropic can go bust, but that's an equity problem; the hyperscalers are the borrowers. The Global Crossing redux: one or two of the five win, because markets tend toward monopoly.

18:22 Slide 12: breakevens say it isn't inflation

• Nominal rates are up 150bp while the 10-year TIPS breakeven closed at 2.34% versus a 2.35% four-year average. "No one cares about inflation… we're looking in the wrong place."
• A 6% deficit with no war, no recession and 4.1–4.2% unemployment is "gross irresponsibility." Rates are pricing the US as a going concern: what yield it takes to hold dollars as a core asset, not bankruptcy risk.

20:41 Is CPI rigged?

• He concedes CPI has been "cooked a little" since the Clinton-era hedonics changes, made to trim Social Security outlays. But the Fed itself steers by CPI, and "I propose the idea of trust. Doesn't mean I'm right."

23:00 A mortgage bond is a covered call

• Model it as buying a cash 10-year at par and selling a 3-year call struck at 105: downside open, upside capped. There's no credit risk in Fannie, Freddie or Ginnie: "gold, guns, and cans of tuna" if they fail.
• Three years ago 71% of the market had coupons of 3.5% or lower, with an average price of $78.94. That was far from the strike, so there was little convexity and spreads were tight.

24:57 Recouponing the stack

• Low coupons prepay (paydown, refinancing, divorce, death, moving) and new production carries higher coupons: over $1T of Fannie 5.5s now. Low coupons are down to 51–52% of the index and 37% has a 5% coupon or higher, all very negatively convex. "An option is most convex when it's at strike."

26:09 Convexity = unbalanced return

• A coin flip paying ±$1 has zero convexity. Make 2 or lose 1 is positive; lose 3 or make 2 is negative. The market prices this: a negatively convex bond needs extra yield over the 5% Treasury, and that extra spread is the mortgage spread.

27:37 MBS spreads widening as the curve flattens

• The par mortgage spread over the cash 10-year has widened from about 95 to 110bp. Two reasons: higher volatility (you're short the option), and more importantly the yield curve. When it flattens or inverts, the option gains value and MBS widen. If it keeps inverting, "it's going to get worse."
• Mortgages are roughly a quarter of issuance, so this will matter for all fixed income, though less than 20 years ago when Fannie and Freddie were "government sponsored hedge funds."

29:09 Index MBS ETFs: fall fast, then track Treasuries

• "Be careful." Big index ETFs "are going to go down quickly, but soon enough they're just going to follow Treasuries": as prices fall, the call moves away from the strike and convexity fades. Prices end much lower, but the fall slows.

31:38 Can AI take all the alpha?

• Models can't predict the future; they optimise portfolios and react to news. Humans will "do just fine" until the future is knowable. If all management went to AI, "there'd be no money to be made at all."
• He agrees with Michael Green's passive-investing critique ("I think Mike's right") and notes Green has a new product in registration.

35:56 "It's always about character… it's never different this time"

• Saddle and buggy-whip makers survived the car. China's WTO entry was a disruption the government didn't help with retraining. He expects AI to become a tool for civilization.

36:53 MMT victory lap

• MMT assumed politicians would raise taxes or cut spending once inflation appeared. "Good luck in prison with that": it ran against the human drive to get re-elected.

38:00 Page 15: his fiscal predictions

• The deficit needn't be solved tomorrow; there needs to be a committed path. His predictions: uncap Social Security (capped at $184k), raise the retirement age slowly (65 in 1936 would equal 84 today), and means-test Social Security and Medicare ("always been a straight up tax").
• Also end step-up basis (defer gains, but tax them), revive some version of Simpson-Bowles, and start passing the "70-30 issues" to rebuild confidence.

39:51 ETFs, part 1: linear vs daily-reset leverage

• There are now more ETFs than stocks. Leverage through futures is linear and fine to buy and hold. Daily-percentage-reset ETFs are fine for a day or a week and "really problematic" after that: +20% then −20% leaves $96, and volatility drag "will eventually burn that ETF to zero."

40:55 ETFs, part 2: return of capital

• A high yield is "eye candy." Check whether the underlying assets earn the distribution. If they over-distribute, the price bleeds month after month as a "self liquidating trade." Many ETFs paying 10%+ likely return capital when junk yields 6–8%.

42:19 ETFs, part 3: the liquidity mismatch

• An ETF is a hybrid, neither open-end nor closed-end. Redeemers get NAV even when the manager sells illiquid underlyings (derivatives, total return swaps) below NAV, which harms the remaining holders.
• Volatility is low (VIX ~17, realized 10–11; MOVE realized ~68). A real vol spike with heavy flows "could be a real problem." Details are in his commentary Looking Under the Hood of ETFs.

44:43 Bitcoin to zero; USD stablecoins a "grand idea"

• Bitcoin is "an active war against a sovereign state" and useless for transactions.
• USD stablecoins create Treasury demand, useful with Japan selling. There's no escape route to a rival stablecoin because no other currency (ruble etc.) is liquid enough. Venmo is already a de facto stablecoin: "I'm not worried at all."

46:04 convexitymaven.com; the cleanest dirty shirt

• About 20 years of free commentaries, plus a podcast, at convexitymaven.com; email him to be added to the list. "The US is still the best place out there": Bill Gross's "cleanest dirty shirt."

47:02 Market Desk (hosts, not Bassman)

• Patrick Ceresna's trade of the week: a TLT Jan-2027 85/77 long strangle (~$1.79) to own bond volatility on Bassman's themes; breakeven needs a ~7% move by expiry.
• Host commentary, not Bassman's: the hike was priced in; the 10-year tests 5%; large specs are long the 2s/5s and short the long end (COTSignal.com). The S&P is near negative gamma with CTA triggers below ~7,500. Brent/WTI are above $100 (Hormuz); the DXY reclaimed 100; gold is testing $4,400; copper spec longs are at a 100 score; nat-gas specs are at a record net short (~220k contracts).