1Iran invasion odds jumped six points in a day, and crude traded lower anyway
The largest move on any market checked today was political, not financial. Polymarket's contract on whether the United States invades Iran before 2027 rose 6.0 points in twenty-four hours to 23.5%, on $1.57m of volume over the same window — the most heavily traded and deepest contract on the board. The trigger is not in dispute: US strikes on Bandar Abbas and a resumed blockade of the Strait of Hormuz.
What did not follow is the barrel. traded 1.63% below the session's open, and Prof G Markets described Brent as volatile rather than climbing while the strikes continued. The adjacent Hormuz contract — whether shipping traffic returns to normal by 31 December — sat at 55.5%, exactly where the previous day's reading left it, on $37k of twenty-four-hour volume against $5.2m lifetime. The venue's own change field reported −3.0 points, which the unchanged price does not support, so this is a level to read rather than a move.
One second-hand datapoint leaned the other way: CNBC, picked up through the day's news pull, reported Kalshi traders pricing petrol above $4 a gallon by the end of July. That was not verified at source. And in posts across 14–16 July, President Trump described oil as "flowing like never before, thanks to the awesome Power of the United States Military" — a claim about abundant supply, made into a market that spent the day repricing the odds of a ground war.
The escalation premium, in other words, is being paid in a political contract and not yet in the price of oil. Either the invasion market is pricing a tail crude has no reason to discount, or crude is late to it.
2Two weeks out, the rate market is arguing about a hike, not a cut
Polymarket's July decision contract prices no change at 95.65%, up half a point, against 3.85% for a 25bp hike and 0.35% for a cut, on roughly $547k of twenty-four-hour volume. The 29–30 July meeting is, as far as that market is concerned, already spoken for. The live question is the rest of the year, and there the contract on a 2026 rate hike sits at 51.5%, up a point — a coin flip on tightening, with easing barely in the frame. It traded only $21k in twenty-four hours against $4.2m lifetime, so the level is the signal and the daily tick is not.
The June data explains the shape of that argument. Consumer prices are up 3.88% year on year even though the seasonally adjusted index fell 0.42% on the month. Producer prices fell 0.28% on the month — the cooler print behind Wednesday's rally — but are still up 5.51% on the year. Unemployment ticked down to 4.2% from 4.3% and payrolls added 57,000. Momentum is cooling; the levels are not. The curve reads it the same way: two-to-ten steepened 6bp to +0.42, while the 10-year eased to 4.58% from 4.62% two sessions earlier.
That combination is what moved metal. Reuters tied gold's drop to rate-hike bets, and the tape matched — finished 2.20% below its open and 0.56% below, on the same day the odds of a US invasion of Iran jumped six points. For gold, the rate channel outran the war channel.
3Central banks kept buying gold while ETF money left, and the price followed the ETFs
Felix, on Felix & Friends, made the official-sector case in a video titled "THEY are preparing for $30,000 Gold." That number appears nowhere in the body — no price target is offered at any point — which is worth knowing before weighing what follows.
What follows is flow. He says the People's Bank of China bought 15 tonnes in a single month, its largest since 2023 and the twentieth consecutive month of buying, on the same day Bloomberg ran a headline declaring gold's bull market over. He puts first-quarter 2026 net central-bank buying at 244 tonnes, naming Poland, Uzbekistan, Kazakhstan and Czechia as buyers and Turkey as a net seller, and cites a survey in which 74% of central banks expect the dollar's share of reserves to fall within five years. He adds that China's largest index fund is now a gold fund at roughly $13bn, ahead of its largest stock ETF at about $12bn.
The other side of that ledger is the exchange-traded money: roughly $18bn has left gold ETFs since the peak, which he reads as late retail capital giving up into a 28–30% correction from gold's 2026 high. Two caveats travel with all of it. None of the figures arrive with a named source or a date, and the size of that drawdown is load-bearing for the whole argument. And Felix runs a paid gold-education business — the video pitches two of his own properties — though he takes no sponsors, names no instrument to buy, explicitly declines to tell anyone to sell everything for gold, and says a stock portfolio including tech remains useful.
So the strongest version of the gold case on offer is a multi-year flow argument with no dated catalyst attached to it, which means a single session can neither confirm nor break it. The session went the other way regardless.
4New York made permits, not chips or power, the binding constraint on AI data centres
Prof G Markets reported the first statewide data-centre moratorium in the country: Governor Hochul's executive order pauses state permits for the largest AI data centres for one year. The context runs wider than one state — roughly $130bn of projects were already delayed or blocked in the first quarter of 2026, and Gallup finds 7 in 10 Americans oppose data centres being built near them, opposition the segment describes as non-partisan and likely to spread.
The counter-argument arrived from the top inside the same forty-eight hours. In posts across 14–16 July, President Trump attacked the New York decision by name, calling data centres "One of the biggest Driving Forces in the Future for Jobs" and "Money Machines for the State." Both sides of that fight are now explicitly on the record, which is new.
Underneath the politics sits the return question, and it surfaced in three separate places. Prof G Markets flagged renting out compute capacity — with named as a counterparty — as a marker of scepticism about what AI capital spending earns. Felix set roughly $700bn of big-tech AI spending against a claim that 95% of firms investing in AI see no positive return. And Equity Mates, whose electrification thesis needs the build-out to continue, named the same dependency as their own key caveat.
Their episode is where the physical bill for all this gets counted: a $500m Microsoft data centre in Chicago consumed 2,177 tons of copper, data-centre energy demand rose 17% in 2025 and is expected to double by 2030, an electric vehicle needs about 70kg of copper, and India's copper consumption is forecast to run from 1.7Mt to 5Mt by 2035. Copper is at an all-time high, and BHP now calls it its highest-conviction long-term commodity and, for the first time, its largest profit contributor. One disclosure belongs with that: Equity Mates Media sits inside the Betashares group, an ETF issuer — disclosed in the outro, but not at the point in the episode where a Betashares fund is discussed.
The constraint on the build-out is therefore migrating from what it costs to what it is permitted to do, and permits are decided locally and slowly. That does not show up in a quarterly print the way a capex cut does.
5Small caps held green while memory and storage took the day's heaviest selling
The day's only fresh hard print was retail sales: advance June sales of $768.6bn against $766.9bn in May, up 0.22% on the month and 6.72% on the year. It fits the consumer-resilience thread running through the day's news, and a CNBC framing of small-cap breadth as the strongest in three decades.
The tape said something similar in miniature. Measured from the session's open, was up 0.22% while was down 0.78%. The heaviest selling clustered in memory and storage — −7.45%, −6.37%, −2.90%, with −1.81% — while set an all-time high on the integration of Alibaba's Qwen model into Apple Intelligence in China.
One measurement caveat applies to every figure above: these are moves from the session's open, not against the previous close. , down 3.98% on that basis, had gapped up around 6% on its results and then faded — a fade off a gap, not a down day. The rotation is also visible in what the day's stock ideas looked like: every single-name idea in the run was a company trading far below its own high.
What the sources recommended
Six single-name ideas came out of six videos, and all six came from one channel — Ross Givens, whose two videos in the run are both pitches for his own $5-a-year paid membership. Only one is a spoken buy; the other five are chart setups carrying entry and invalidation levels rather than stated recommendations.
The spoken buy is , Stereotaxis — a roughly $160m magnetic surgical-robotics company trading near $1.65, about 55% below its late-2025 high, on a support zone that has held since 2019. Givens' case is four dated events: FDA approval of the Genesis X system in November, the first that needs no reinforced-room construction; clearance of Synchry in April; the first US procedures with the MAGiC disposable catheter in April; and the 9 July close of French acquisition Robocath, adding coronary and neurovascular robotics. Revenue is $32.4m, up about 20%, against a roughly $23m net loss, and four covering analysts all rate it a buy at an average target of $4.13 — about 150% above the price. He volunteers the bear case unprompted: "win big or fail big", likely dilutive raises, systems that still have to ship, integration work ahead. What he does not say is whether he owns it, and on a $160m company that omission is the material one. The $4.13 average rests on four analysts covering a microcap, and the $20–30bn addressable market quoted is the company's own sizing, repeated as given.
The five setups — , , , and — come from a manual screen of roughly 600 S&P 500 and Nasdaq 100 names for stocks down 40–50% or more that have based and reclaimed their 30-week moving average, with an earnings and sales growth filter on top. (Klarna) is the one he pushes hardest: a September 2025 listing that fell from about 55 to 12 and is now breaking out, entry around 19.50, invalidation below 17, target in the mid-30s. reclaimed the same average after a roughly 60% drawdown; fell from 77 to 30 before basing; went from 160 to 85 and trades near 111, which he calls a slower mover. he flags as possibly early — he wants a close above 200 on volume first, so on his own framing it is a watch and not an entry.
Two features of the method travel with those names. The video's title promises that "the last time this happened" stocks exploded, but no dated precedent, backtest or hit rate is given anywhere — the setup is recurring, not historical. And a filter that selects companies 40–50% off their highs which have only just reclaimed a long moving average is, by construction, a falling-knife screen; no reversal base rate is offered next to it.
The one avoid-shaped remark in the run came from Ren on Equity Mates, who said he would personally steer clear of the battery-supply-chain ETF route in favour of exposure to the core minerals themselves — and the fund in question is Australian-listed, so it is not a US instrument at all. On the US side of that theme, only Vale had its ticker spoken aloud; Albemarle , Quanta Services , Freeport and were named as companies rather than symbols. Both hosts describe the theme as a commodity cycle rather than a hold-forever position, and the trailing-year moves they quote — Pilbara up 381%, Liontown up 275%, MinRes up 216% — describe gains already made.
Nothing before 29 July can settle the hike question
The next Fed decision lands 29–30 July, and at a 95.65% hold it is already accounted for, which means the 51.5% on a 2026 hike gets moved by data rather than by the meeting. Between here and there, the inflation and labour prints are the only things with the standing to shift it.
The first section offers a cleaner test. The invasion contract is liquid enough that its next move carries information: either crude begins tracking the 23.5%, or the 23.5% comes back down. The Hormuz normalisation market, at 55.5% for year-end, is the slower version of the same question.
And on the day's only spoken buy, the proof point is mechanical rather than narrative — whether Genesis X systems actually ship in volume and the Robocath acquisition converts, measured against a $23m annual loss on a $32.4m revenue base.
