On October 9th Mohammed Al Jaber, Saudi Arabia's ambassador to Yemen, declared that the Bab el-Mandeb had been “cleansed of the remnants of the terrorist Houthi militia”. A Houthi spokesman replied that the strait was 'pure from the filth' of the Saudis.

Both sides agree that the water is clean.

What they cannot agree on is who is standing on the rock in the middle of it.

The rock is Perim, 13 square kilometres of volcanic scrub that splits the strait's 28km mouth. The Houthis landed on it unopposed on September 11th, the day after government troops pulled out. Polymarket gives a 51% chance that it is out of their hands by New Year. On October 10th, the day after the ambassador spoke, that figure touched 81%.

Straits matter in proportion to what has nowhere else to go, and this year that is a lot. With Hormuz largely shut since March, Saudi Arabia has been piping about 4m barrels a day of crude west to Yanbu, on the Red Sea, and much of it then sails south through the Bab el-Mandeb. The Houthis have blockaded Saudi shipping since July 20th. A week's war-risk cover for a tanker calling at Yanbu now costs about $3m, against roughly $100,000 before the fighting.

The Saudi answer, launched on October 4th and named Dawn of Yemen, sent the Giants Brigades, a southern force, up the coast under what the coalition says were 100 jets. Traders, who had put the island's recapture by New Year at 11% at the start of the month, began to buy. On October 9th the Giants released footage of their men searching the island, the ambassador made his announcement, and the price ran.

Then came the doubts. A government military source told AFP that the fighting was fierce and 'advances are limited'. On October 11th the Washington Post reported that control of the island 'remains disputed'. None of Reuters, AP or AFP has independently confirmed who holds it. The price fell 30 points in two days.

The contract explains why that matters. Its rules say that a 'claim that the Houthis have lost control will not qualify', and that if control is 'contested, unclear, disputed' on the deadline, it resolves No. The question being traded is therefore less who wins the island than whether anyone credible can say so. The October contract makes the point. If traders believed the ambassador, it would be trading near certainty. It is at 27%.

The Saudi case is still decent. The kingdom has the aircraft, the money and growing American help: aerial refuelling, more than 200 American analysts in its command centres, and a president who, asked on October 10th whether he would join the strikes, said: 'We may. We're going to look at it.' Islands are easy to lose, as government troops showed in September, and coalition-backed forces have taken this one back from the Houthis before, in October 2015.

The Houthis do not need to hold Perim to win this bet. On October 10th, the very day traders were most confident, a Houthi missile hit King Khalid airport in Riyadh and killed at least 12 people. A movement that can reach the Saudi capital can keep a small island contested for a long time, and contested is all it takes.

We think traders were right to sell the ambassador's claim, and that the year-end price still leans on Saudi communiqués more than it should. Until a news agency can say who holds the island, the contract is a bet on access as much as on arms.

In the 1920s Perim's coaling station loaded more coal than Aden. Its chief export these days is communiqués. Traders, sensibly, are paying about half price for them.

Polymarket's market on whether America confirms that aliens exist has traded $73m, and for most of its life the bet has been about UFOs. Donald Trump ordered the government's files released in February, and the price peaked at 20.5% in April. Six batches of Pentagon documents later, nobody has produced a spaceship. The price hit 2.8% on October 3rd and stands at 4%.

The stretch is that traders have been watching the wrong agency. The rules never mention UFOs. The market resolves Yes if the president, a cabinet member, a member of the Joint Chiefs 'or any US federal agency definitively states that extraterrestrial life or technology exists'. Life includes microbes. NASA is a federal agency, and its administrator, Jared Isaacman, said in June that samples from Mars had a 'very high probability' of showing microbial life.

The obstacle is 'definitively'. The samples are still on Mars, and scientists hate the word. But Mr Isaacman, a payments entrepreneur who has commanded two private space flights, is not a cautious career scientist. Nor has the UFO route closed. A Pentagon waiver in September freed officials to set aside their non-disclosure agreements and tell the government's new UFO archive what they know, and a disclosure amendment to the defence bill awaits the lame-duck Congress.

Still a stretch, but the price buys one definitive sentence from any one of dozens of agencies before New Year. Nobody needs to land a flying saucer on the White House lawn. Somebody at NASA just needs to stop hedging.

Polymarket has settled four contracts on a military clash between NATO and Russia, and all four paid out No. The one that runs to New Year rose four points on October 12th, to 25.5%, on the day General Alexus Grynkewich, NATO's top commander in Europe, told a security forum in Warsaw that 'if someone is foolish enough to attack us, we will fight'. No incident has been reported to explain the move.

The year has been tense. Allied jets have shot down about ten drones over NATO territory in 2026, and Romania has logged at least 23 airspace violations this year, more than in the previous four combined. In September a Russian frigate fired flares at a Danish military helicopter.

None of it counts. The contract demands 'the use of force' between the two sides' armed forces. Its rules exclude airspace violations, warning shots, cyberattacks and the downing of one-way attack drones aimed at a third country, which describes nearly every allied shootdown so far. The last event that would plainly have qualified was Turkey's downing of a Russian Su-24 in 2015.

Our analysis puts it around 10%.

The market's prices imply roughly a 10% chance of a clash every month: the rate of a war that keeps nearly starting. The record since 2015 implies a fraction of that. Our figure, about 4% a month, gives this year's escalation its due, and the grey zone too. The rules count shooting down drones 'which are not munitions', and unarmed Russian decoys have strayed into Poland and Lithuania. The August contract was disputed before it settled No, so somebody will try again.

What changes things: an allied jet downs a Russian reconnaissance drone, an Orlan 10 say, over NATO soil, or Russian forces fire on a NATO aircraft or ship. Either would settle it the same day, and 25.5% would then look cheap.

Polymarket gives François Hollande a 3% chance of winning France's presidency next spring. That looks generous for a man whose approval rating hit 4% in 2016, the year he became the first president of the Fifth Republic not to seek a second term. It fell two points this week, after Raphaël Glucksmann won the centre-left primary that Mr Hollande sat out. Here is the case that it is too low.

Start with the round that decides it. Marine Le Pen, free to stand after the Paris appeal court ruled in July, leads every recent poll and is a 42% favourite. In Ifop's head-to-heads in August she beat Mr Hollande by 54 to 46. That is a defeat. But it is the same score as Gabriel Attal's, better than Mr Glucksmann's 45 and far better than Jean-Luc Mélenchon's 33. Only Édouard Philippe did better, at 48, and by September Toluna had him losing 57 to 43.

Then look at Ms Le Pen's calendar. The Cour de cassation, France's top court, will rule on her appeal by early April, a fortnight before the first round on April 18th. If it rejects the appeal, her electronic tag could apply in the middle of the campaign, and some lawyers argue, though others disagree, that the appeal has revived an earlier ban on her standing at all. Her obvious stand-in, Jordan Bardella, trades below 1%.

Getting to the runoff is the hard part. Mr Hollande polled 12% in Ifop's first-round survey in August and 8% in September, well short of the 21% or so that second place currently takes. But the centre is split: Mr Philippe falls to 16% when Mr Attal runs too. A wounded Le Pen and a divided centre is the scenario in which the left's most electable veteran stops looking absurd. Mr Hollande, who will decide by mid-December, told La Repubblica that a run is 'not a decision to be taken lightly'.

The case against is simple. Mr Glucksmann won his primary outright and now owns the lane, and most of the 3% is a bet that Mr Hollande runs at all. In 2012 he promised to be a 'normal president'. In a race this abnormal, normal may yet sell.

We’re up

A real $10,000 book on Polymarket. See all of our current positions - @thebasementview.

Yes on Likud winning 20 to 24 seats: 750 shares at 43%. Cost: $322.45 plus a $7.35 fee.

Rationale: Israel votes on October 27th, and the polls cannot agree on Likud. Eight surveys published between October 13th and 16th gave it anything from 18 to 29 seats: the six mainstream pollsters 18 to 21, Channel 14 and i24 29 and 27. The middle of that spread sits in the 20 to 24 band, and Likud has a habit of beating its final polls; in 2015 they gave it 21 or 22 seats and it won 30. Polymarket prices the band at 42%. We think it is nearer 50%.

Expected return: If Likud lands on 20 to 24 seats we collect $750, a profit of $420.20 after fees. Otherwise we lose $329.80. At a 50% chance that is worth about $45, roughly 14%. It only loses money on average if the true chance is below 44%.

Risks: The polls are spread across 11 seats, and the mainstream ones are drifting down: the under-20 band rose 11 points this week after Channel 13, Channel 16 and Maariv all put Likud on 18. Small parties hovering around the 3.25% threshold can also move several seats at once on the night.

Nothing to sell yet.

Know one person who is fed up with traditional news? Forward them this.

Been forwarded this? The button gets you your own copy every Sunday.