Nassim Taleb On XIV: “Why People Who Make Money Are Usually Wrong”

Original post

Echoing Mark Spitznagel’s insights into how ‘naiveté’ led to the epic losses experienced by many ‘nickel-picker-uppers’ this week in the short-vol game, Nassim Taleb takes to YouTube to provide some more color on the fallacy of forecasting and what destroyed XIV traders.

Taleb begins by noting that “many people attempted to profit by forecasting volatility [would drop] and from the fact that  the contract [in this case XIV] was poorly constructed… they were right, until they were destroyed.”

Academics cannot get the idea that you don’t have to be right about the world to make money.

“Antifragile explains why understanding x is different from f(x) the payoff or exposure from x. Most of the harm/gains come from f(x) being convex or concave, not from understanding x.

Forecasting is off an average, and average is for academics and other morons.”
As Valuewalk’s Jacob Wolinsky writes, this video illustrates the point with XIV that went bust while being correct about volatility –and why people who make money are usually wrong.

[embedded content]

Transcript:

Friends. Let’s discuss the problem forecasting one same time explaining the XIV/VIX trade. A bunch of people constructed the contract saying the VIX is overpriced, mis-prices volatility, how over-forecast the variation in the market is; the contract is poorly designed, so let’s make money off of that… And they were right but they were destroyed and lost all of this money.

Why?

Because they didn’t realize that forecasting has nothing to do with P&L, nothing to do with real-life; what matters is you payoff function.

So these people were right. If you invested say $25 you would have made some money. I think a high of $146 overtime. And in one day they lost everything.

It’s now at $5 and probably shopuld be lower.

So what what did they do that was wrong.

What they did is not understand that being right on a random variable X doesn’t mean making money out of it. Your payoff f(x) needs to be aligned with what you’re forecasting. And in fact the payoff function f(x) is never X; and f(x) can be very complicated. In real-life, most people think you’ve got to focus on X or academics or other idiots.

f(x) is what you focus on when you make the decision. It’s much easier to understand your function of a random variable than the variable itself. That’s what I said throughout ‘Antifragile’, and very few people are getting it.

[00:01:57] Academics cannot get the idea that you don’t have to be right about the world. You have to make decisions that are convex, in other words, decisions that make sense and in fact you don’t have to be right about the world. And this also explains why paranoia is entirely justified, if you’re f(x) is concave.

So your overpricing and underpricing of probabilities is not what matters; what matters is your payoff function.

So let’s see what happened here… Let’s take your payoff function. Remember you have a payoff function that’s a polynomial – you know pretty much everything in life is some nonlinear function that can be expressed through splined polynomials (and of course you can do it a more sophisticated way, but that’s pretty much what it is).

This is a way for us to understand first-order non-linearity.

When you plot (1-x^2), you see what happens here (below) and so I build a forecasting function that is the mean of (1-x^2). So what does it do here

[Taleb shows the function’s attempts to ‘forecast’ the 1-x^2 using its historical mean to that point]

[00:03:45] Although you’re right on X you’re wrong in your payoff function you’re going to be harmed big time.

Now let’s take an extreme case.

This is why I often say, “I’ve never seen a rich forecaster; good forecasters are always poor because they don’t forget that the average forecast is not what matters. What matters is not to be harmed by these concavities.”