The 2008 machine, rebuilt responsibly. On sports gambling.

ELI5

System is now live. Watch the active monitor →

Everyone is arguing about France at 16 cents

On June 11, 2026, the largest sporting event in human history kicks off in Mexico City, and for the first time, the dominant way a large slice of the world bets on it is not a bookmaker. It is a prediction market — a venue where you buy a contract that pays one dollar if a thing happens and zero dollars if it doesn't, at a price that floats between zero and one hundred cents and is the market's live probability of the thing.

One World Cup contract has turned over $1.6 billion before a single ball was kicked. Sports are now roughly 39% of the largest venue's volume. Combined monthly volume across the major venues went from under $5 billion to roughly $24 billion in eight months. This is, functionally, a parallel derivatives exchange for reality, and it has discovered sports at the exact moment sports has discovered it.

Everyone trading these contracts is arguing about whether France is overpriced at 16 cents. Everyone trading these contracts is missing what the contract is.

A binary claim is a zero-coupon bond with a coin flip attached

A "Yes" share on Brazil wins the World Cup pays $1 if Brazil wins and $0 otherwise, and trades today at about nine cents. A fixed-income desk would not call that a bet. They would call it a zero-coupon, zero-recovery bond with a single binary credit event at maturity, where the credit event is "Brazil fails to win the World Cup," and they would not be being cute. The payoff structures are identical. The nine cents decomposes exactly like a bond price. There is a yield, a duration, and a hazard rate, and the hazard rate is Croatia.

binary claim ≟ zero-coupon bondFIG

And like any bond, it has a variance — except here the variance requires no model, no history, and no estimation. It is p(1−p), arithmetic, exact. Highest when the contract trades at 50 cents, collapsing to zero as the outcome resolves. Uncertainty is mechanically maximal at kickoff and decays to nothing at the final whistle.

The index fund, because diversification is just a weighted sum

Nobody wants to click 48 times to express "the favorites will broadly be fine," so the first product builds itself: pool the favorites' contracts, issue one fungible token against the basket, collect a small wrapper fee on creation and redemption. An ETF on national footballing competence. The portfolio math is the ordinary kind — a weighted sum of variances, plus the cross-terms, every pairwise covariance between the constituents.

Var(portfolio) decompositionFIG

In a normal tournament the cross-terms are nearly zero. Brazil winning its group in one city has little to do with Morocco winning its group four time zones away. The games look independent, the covariances vanish, and the basket feels safe.

File that feeling of safety. It is the exact feeling that was mispriced in 2007.

The tranche, or: how to turn 48 soccer games into a CDO

Different buyers want different risk, and traditional finance solved that with a waterfall: pool the cashflows, then pay them out in strict order of seniority. The equity tranche eats the first upsets and collects the fattest yield — these are the degenerates, and they know it, and they are being paid for it. The mezzanine sits above them, untouched until the equity buffer is gone. The senior tranche sits on top and only loses if the upsets chew through everyone below it, which, in a normal tournament, essentially never happens. So it pays a low yield and is marketed — sold, believed — to be safe.

You have just built a collateralized debt obligation out of the World Cup. The obligors are national teams. A default is an upset. Recovery is zero. The attachment points are contract parameters. And the senior tranche is safe for exactly as long as the upsets stay uncorrelated.

If favorites fail independently — one here, one there, scattered randomly across the bracket — the total number of upsets clusters tightly around its expected value, the junior buffer absorbs them, and the senior tranche is genuinely money-good. Sums of independent variables concentrate; the probability of many simultaneous failures is a product of many small numbers. But if some single factor causes many favorites to fail at once, the distribution grows a fat upper tail, and the tail goes straight through the equity, through the mezzanine, and into the tranche everyone was told was safe.

The senior tranches of subprime CDOs were rated AAA on models that assumed mortgage defaults in Florida were roughly independent of defaults in Nevada. They shared a common factor. When it turned, correlation went to one, the tail the models said was empty turned out to be where all the mass was, and "safe" went from par to pennies. The instrument didn't lie. The correlation assumption did.

Why this World Cup's correlation structure is uniquely radioactive

So everything depends on one question: what could make a dozen favorites fail at once? In most years, nothing — football is noisy and local, and independence is a decent approximation. A World Cup CDO in 2014 would have been a boring, well-behaved product. 2026 is not most years. 2026 has common factors, they are not hypothetical, and they are in the news today.

common factor → correlated defaultsFIG

The senior tranche of the World Cup CDO is short exactly this: a correlation spike that can arrive on the strength of a press release. It is short correlation, and it doesn't know it. Which raises the question of who sells it the protection it needs.

The synthetic VIX, or: volatility as a tradeable asset

Now stop watching prices and start watching how violently they move. Sum the squared price changes across the basket — every upset, every VAR reversal in the 89th minute, every favorite suddenly on the brink — and you have a live, continuously observable index of how much chaos the tournament is producing. Package it as a token and you have built a volatility index on a soccer tournament. A Sports VIX. (You said you'd remember the shape from § II. This is where you buy it.)

Why is this more than a casino game with a Greek letter painted on it? Because the equity VIX is useful for exactly one reason: it spikes precisely when the market craters, so long volatility offsets long equity by construction. And our version has the same property, mechanically. The event that does maximum damage to the senior tranche — a wave of correlated upsets — is a cluster of contracts violently repricing from near-one to zero, which is the same physical event that sends realized variance through the roof. The hedge isn't statistical. It's mechanical. Same coin flips, two functions.

senior tranche ⟷ variance tokenFIG

The correlation note: the instrument that pays when the world stops being independent

One instrument left, and it's the sharp one: a contract that doesn't bet on upsets, or on chaos, but on whether bad things happen together. Take the twelve favorites' entry prices. If the games are independent, the number of upsets K follows a distribution you can compute exactly from those prices — the Poisson–binomial, mean Σ(1−pᵢ), variance Σpᵢ(1−pᵢ), no model, no history, nothing to argue about. Set a threshold K* at its 95th percentile and write the contract: pays $1 if K ≥ K*.

Under true independence this resolves Yes almost exactly 5% of the time, by construction, and should trade near five cents. It pays off dramatically more often than that under exactly one condition: the upsets are positively correlated. You are not betting on upsets. You are not betting on volatility. You are betting that the independence assumption breaks.

And the alpha lives in one place: the five-cent baseline is only fair if the games are actually independent when you strike the contract. If a common factor is already visible — the travel-ban news is already breaking, three southern venues are already forecast for 40°C — then the true probability of K ≥ K* is far above 5%, and the contract is mispriced at birth by everyone who computed the lazy independent baseline. The crowd prices the bracket as 48 independent coin flips. It isn't. It never was. In 2026 least of all.

Yes, we noticed

At this point you have noticed that we have reconstructed, instrument for instrument, the architecture that brought the global financial system to its knees eighteen years ago, and you may be wondering whether we are aware of this. We are. It's load-bearing.

Here is the comparison, stated as mechanics rather than morals:

The 2008 complex failed because the safest collateral in history turned out to have a common factor, and the structure around it was leveraged, opaque, and systemic, so the surprise was fatal. This complex starts from collateral nobody would dream of calling safe, funds every position in full, and publishes its one fatal assumption as the centerpiece of its own marketing. We are not claiming this says something reassuring about us. We are noting that it says something extremely uncomfortable about 2008.

The part where it stops being a metaphor

Everything above would be a creative-writing exercise except for one technical fact: a position on the largest prediction market is a standard token, on a public chain, in a wallet you control, fully collateralized one-to-one by a regulated stablecoin — and as of two months ago, the exchange lets smart contracts trade it directly. The plumbing for "a contract that holds and trades prediction-market positions" already shipped. We didn't build the rails. We just read the documentation.

on-chain substrate · CTF / pUSDFIG

Outcomes resolve through an optimistic oracle: a proposer posts the result with a bond, a dispute window opens, and if nobody objects, winning tokens redeem for a dollar a few hours after the final whistle. In the gap between "mathematically decided" and "redeemable," winning shares trade at 99.5 to 99.9 cents — the missing fraction is the market pricing the cost of waiting for certainty, visible in the order book. Even the boring plumbing here generates a yield curve.

optimistic oracle · resolution flowFIG

The build order is the order you just read. Each module is independently useful, and each one is a real product even if the next never ships:

Nothing in the stack is about soccer. The instruments require exactly one thing from their underlying: binary contracts that resolve through a trusted oracle. Elections qualify. Fed decisions qualify. Rocket launches qualify. The World Cup is simply the loudest possible place to plant the flag — the first underlying, not the last. Hence the name.

Satire or spec?

The teams are real, the contracts are real, the oracle is real, the $1.6 billion is real, the math is exact, and every component compiles. The only thing that decides which one it is, is whether someone builds it. The tournament starts in —. We have to decide too.

Reality-Backed Securities is brought to you by Jaime Rogozinski, who previously brought you WallStreetBets, and who notes that the last thing he built also started as a joke.