This is probably because the economist who first looked into relationship between the yield curve and recessions used 10y-3m [0]. His reasoning is that the 3m gives a better outlook of the conditions happening at the moment than the 2y.
I don't know if $45B is a line item to them, but to give you a sense of scale of how much money the Saudis have, Aramco had a net income of about $34B in the first half of 2017 [0].
In Peak: Secrets from the New Science of Expertise, Ericsson had chess grandmasters look chess boards of real games, and they could recreate the board in a pretty short fashion.
But when they took chess boards of randomly placed pieces, they could not remember the placement of pieces any better than a random group of people.
The idea is that through practice, you develop the ability pattern match and chunk together the data into more cohesive blocks, and memory fewer of these blocks.
The rough analogy in software is to take a quick look at some code and recognize the design patterns, and remember the design patterns used rather each individual character.
In fact, words are a form of chunking -- we don't remember hundreds of character sequences, but we remember the words instead.
Matt Levine wrote an article (with charts!) on VIX[0] back in 2014 that basically says the same thing: VIX is more of a measure of past volatility than an indicator of future volatility.
> 1. Crime pays less than it used to. Back in the good old days, people carried more cash on them. Pickpocketing and mugging was reasonably profitable when everyone had money. These days, everyone has credit cards. I frequently carry less than forty dollars; I've gone weeks where I have zero cash in my wallet.
On the other hand, now everyone has a smartphone in their pocket. My friends and I entered Southeast Asia with four phones and left with one. None of our wallets were pickpocketed, but that may just be chance.
So the direction may not be that interesting!