A programmed world (the death of predicition markets)

Predicting the World Cup and Beyond

Writing this essay just days after EA Sports predicted the FIFA World Cup winner for 5 times in a row now, starting in 2010. Funny enough, it's a full-circle moment when they predicted Spain in 2010 and now in 2026. But it is insane that a video game production company can rightly call the winner of the world's biggest sporting event for 16 years now. (EA Sports prediction, Spain’s win)

Getting Infinitely Smarter

And it got me thinking, EA FC (previously known as FIFA, FIFA) will only get more realistic to play, and its simulations will only get smarter from here. All they are doing is predicting a tournament of the same 90 minute games. So what is going to stop them in predicting the winner in the future editions?

Betting on the Future

This comes from me spending almost an hour convincing my friend to parlay on Spain winning the World Cup on Kalshi during round of 32s itself, right when the French and Argentine teams were looking unstoppable. Spain drew their first match with a country so small and was not even supposed to be in the world if it wasn't for the 32 to 48 country expansion. But I had him bet 20 bucks on them, and he now had an upwards of $220 in profit purely off of that. I just couldn't stop saying to myself "There is no way EA Sports can be wrong this time".

The Predictable Future

Now that we can predict sport, predicting for all other outcome based events will also move towards this where a barrage of parameters and factors can be simulated to predict its outcomes. Especially political events like presidential and senator elections, the passing of a bill, etc.

Everything is Computable

But think about it, every minute action that influences a bigger action which then influences a bigger action to the outcome of an event can all be computed with basic conditional probability, and then predicted through first principles. When predictive models get much smarter, can take in more data and have access to much more compute, the whole world becomes predictable.

Think about it. If you told someone in the 1700s that weather could become predictable, it would generate the same reaction as yours reading this essay as I make the claim that any outcome can become predictable. But as long as you accumulate data, process it through a competent enough model, and have access to enough compute, you can make a very reliable prediction. Being contrarian would essentially die then, because you KNOW the model is right. So betting on the contrarion opinion would just end up being a dumb decision.

The End of Prediction Markets

Why is why I also claim prediction markets will die soon. It's funny because they are relatively new to world, but will also cease to exist, because their whole base is built on having people bet on outcomes by making an educated guess, not a computed one.

Replicating the World Digitally

But on the bright side, there is a side effect to this. Making the world predictable means you can replicate the world as well. EA FC is so realistic now, all influenced by OVR stats of real-world performance of players, the in-game models influenced by real-world physiques, and even moments and managerial strategy all influenced by real world. When their digital sandboxes to predict events heavily pull from real-world data, not only can they predict right, but essentially create a virtual model that is very loyal to its real-world counterpart.

This is why I also tweeted that EA sports is the most ideal company world model labs to poach engineers from. And not only EA sports, I think video game companies are one of the perfect places from where you can steal some of the best engineers for world model building.

---