1The AI-agent sell-off in travel and back-office stocks looks like basket selling, and Friday's 4% bounce is the first test of that
Consumer AI assistants that can book trips and cancel subscriptions have been treated as a threat to a wide set of companies this month. Jonathan Boyar, of the Boyar Value Group, argued on The Compound and Friends that most of the damage is mechanical. His case: a Goldman Sachs "consumer inertia" basket is being sold as a unit, so stocks fall together whether or not an agent can actually replace them. Michael Batnick, the show's co-host, gave as an example. It fell 7.2% in two days, although its readers are loyal and the business is growing.
Boyar's travel case is specific. fell from about $215 to $157 in September, a 27% drop. He argued an assistant has no hotel inventory of its own, and that small hotels deal with Booking directly. He also noted that ChatGPT launched instant checkout in September 2025 and pulled back by March. He called the sell-off "even more ridiculous". He made the same argument for , which handles about 80% of US proxy votes and is down from $280 to about $160, on fears that tokenised shares would bypass it.
The market moved his way the day after the taping. closed Friday up 4.1% at $163.89, and rose 4.0% to $157.42. rose 3.7% on a Copilot overhaul. The two hosts disagree on what ends the de-rating. Batnick expects a single OpenAI partnership to reset the story, as a similar deal did for Salesforce. Boyar expects no single catalyst and said lower oil or a settled Iran conflict would help most. Batnick also warned that 15 times earnings is not necessarily a floor.
A wider fact sits behind this. Batnick cited a Goldman most-shorted basket up 40% this year, while a long-quality, short-junk trade is down 23%. is down 29% in 144 days, a drawdown he said is rare outside 2020 and the dot-com bust. Boyar blamed multiple compression in every sector except staples on the 5% 10-year, AI fears and a lingering Covid hangover. He noted stocks did fine with a 5% 10-year in the 1990s.
So what: if the selling is mechanical, the bounce should hold without company news. A second leg down on no news would point to a real change in how these businesses are valued.
2Gold held up through the first Fed hike since 2023, and central-bank buying is the case for why
The Fed raised its target range to 3.75–4.00% last week, its first increase since 2023. A higher cash yield should hurt gold, which pays nothing. Ross Givens, on his own channel, said gold futures closed up 1.25% that day instead. He attributed it to buyers who do not trade on rates. By his account, China's central bank has bought for 22 straight months, including 20 tonnes in August. He said central banks bought a record 289 tonnes in the second quarter and gold ETFs have logged eight straight days of inflows.
His case for miners is that costs lag the gold price. He said miners trade at about 15 times earnings, with all-in costs below 40% of the gold price, down from 64% five years ago. He named as the simplest way to own the group and noted that it moves 1.5 to 3 times gold, in both directions. His falsifier is stated: a break below $4,000 an ounce, this year's low, would mean "the thesis is wrong". He compared the setup to 1976, when a mid-bull-market pullback preceded an eightfold rise. The video also sells a paid membership twice, and no counter-case is put to him.
Friday's prices were mild. rose 0.4% and rose 0.6% to $92.89, about 18% below its January high near $113. Polymarket gives gold a 39% chance of reaching $5,000 by year-end, up 12.5 points in a day. That market is thin, with about $2,700 traded in 24 hours, so the move is weak evidence.
So what: gold is rising against higher real yields, not with them. Givens's account holds only as long as official buying keeps outpacing the pull of a 5% Treasury.
3October hike odds eased two to three points while the long end kept steepening
Kalshi prices a 25bp October hike at 64 cents, down from 67 cents, on 535,000 contracts of open interest. Polymarket shows 64.5%, down 2 points, on $2.4m of liquidity. A third hike in 2026 slipped to 38.6% on Polymarket, from 41% a day earlier. A second, smaller Polymarket contract still prices another hike this year at 90.5%.
The long end did not ease with them. FRED shows the 10-year at 5.18% on 24 September, up from 4.96% two sessions earlier. The 30-year printed 5.47%, and CNBC reported the 10-year at its highest level since 2007. The 10-year minus 2-year spread widened to 0.36 points on 25 September, from 0.26 two days earlier. A steeper curve with slightly lower hike odds means term premium, not the Fed, is doing the lifting. fell a further 0.1% on Friday.
Ben Carlson, on a short Compound segment, drew the household consequence. With government bonds near 5%, paying off a sub-3% mortgage early gives up yield. Carlson called low fixed-rate mortgages one of the best inflation hedges available.
So what: the odds of a Fed hike are easing, but long rates are not. The next data prints decide which of the two is right.
4A Hormuz reopening lost odds even as oil fell, and the two have not been reconciled
Reuters reported, citing the Wall Street Journal, that President Trump rejected an Iranian peace proposal. Iran's foreign minister, Abbas Araqchi, said it is now up to Washington to accept a seven-day plan, and a second Iranian official ruled out nuclear concessions. Prediction markets marked the peace odds down. Polymarket prices Hormuz traffic returning to normal by 31 December at 21.5%, down 2 points, on $330,000 of 24-hour volume. By 31 October it prices 5.5%, down 3 points. A final nuclear deal by year-end fell 3 points to 9.5%.
Oil went the other way. fell 3.1% on Friday, after Reuters reported a 2% overnight slide on US-Iran talk of "a path out of war". The EU warned members of an energy-price crisis and asked them to consider curbing demand. On a Prof G Markets clip from an earlier interview, Daniel Baer of the Carnegie Endowment argued crude stays above pre-war levels even with a deal, because of insurance and tolling costs. He also pointed to Black Sea wheat exports that start next month.
So what: oil priced de-escalation while the reopening markets priced less of it. This is unresolved. The next headline from Washington on the seven-day plan settles which one was early.
What the sources recommended
The day's single-name calls came from two people, and most came from one. Jonathan Boyar, on The Compound, accounts for nine of the ten buys. Several of those are long-held positions he reiterated, not new ideas, and his firm sells the research behind them.
Boyar's fresh names were , and , all framed as sold for the wrong reason. He added , now about 15 times earnings, which is what it traded at in 2009 and after 9/11. He added the caveat that pool construction may not have bottomed. On he argued Burger King is halfway through a turnaround that McDonald's is only starting, with a 3–4% yield while holders wait. On his view is conditional on a bid. The stock fell about 11% on Thursday after Barry Diller withdrew his offer, and Boyar expects him to return. He also pointed to Japan's only casino licence, due in 2031. He reiterated , and as positions he "still loves". Batnick added that Uber's CEO and CFO both bought shares recently, and the stock fell the next day.
Boyar's one avoid was . He said he has "passed on" it, and Batnick added that even after five years of falling earnings it "still might not be a buy". Boyar was lukewarm on , calling it "probably less interesting than the others".
Ross Givens's one call was , with the $4,000 gold level as his stated exit condition.
Next week's inflation and jobs data will test the rate claims
Personal income and PCE inflation print on 30 September, alongside revised GDP. September payrolls follow on 2 October. A hot PCE would support the long end over the easing hike odds. reports after the close on 30 September, the first large AI-hardware print since rates moved this high. reports on 28 September and will show whether the business matches the stock's de-rating. Next week's news on the seven-day Iran plan will settle the oil-versus-odds gap.
