Nassim Taleb’s new paper: A Map and Simple Heuristic to Detect Fragility, Antifragility, and Model Error
This is a fun paper which I like a lot as it implicitly tries to get at stability of assumptions and process by ignoring the flaws in assumptions about primary statistical moments used in most stochastic models (i.e. the moments being fixed or measured correctly). True confession, I am interested in intrinsic value not price, but like the paper for what it says about most systems which are by definition complex and wrongly modelled and managed using Gaussian tools and mis-measurements.
One of my favorite parts in the paper is the concept of a random process not equating to the square of its outcomes. :
“For example, take a conventional die (six sides) and consider a payoff equal to the number it lands on. The expected (average) payoff is = 3.5 . Now consider that we get the squared payoff,= 12.25, so, since squaring is a convex
function, the average of a square payoff is higher than the square of the average payoff.”
This struck me as interesting becuase squaring something reflects a proxy of a growth process (albiet) with a single step. Many growth/decay (exponential) processes such as companies or sectors experience are by definition exponential relative to their underlying fundamental drivers. The example is a simple way of showing the miss-fit of a simple stochastic metric to a growth dynamic. It doesn’t even begin to touch on the failure of most models to consider rational boundaries or limits to growth which may be related to fundamental dynamics feeding the process.
Another unquestioned component is that the models are all looking at and using price which is purely an abstracted guess at value being made by traders who don’t understand intrinsic value. My argument is that price is the outcome of two opinions shared at a point in time and both of them are wrong 99% of the time. If the arbiters of price were correct they could beat the market.
The paper is one of Taleb’s more interesting and presents a few concepts as words which could be papers in thier own right: Philostochaticity: Biological, Economic, and Political Systems Starved of Variations and the relationship between convexity and anti-fragility, which some deem to be robustness.
There is a lot in the paper and it is definitely worth more than one read. I am less interested in this for its “price” arguments as I care about intrinsic value, but am more interested in what it has to say about understanding systems (all of which are complex) with out-moded variants on guassian tools and assumptions. Garch just doesn’t work so hot in the real connected world convex world.