Real World Risk Institute Day 1
Take risks you understand rather than trying to understand the risks you're already taking.
Day 1 started out rather nonchalantly. No real introductions, bragging, laurels, or “let’s go around the room and say where everyone is from”. Yes, some jovial poking between Raphael and Nassim while he tried to remember how to use powerpoint (he only uses powerpoint once a year, and only for teaching this class, he insisted). Instead, to quickly get a feel for who is in the Zoom room, Nassim asked us to raise our emoji hands - “How many IT people”? Many. “Analysts”? Also many. “MDs?” Just a couple. “worse - Lawyers?” also a few. “and worse yet… consultants?” about a dozen hands went up. The crowd is much more mixed in RWRI 19, some veteran attendees noted, than it had been in the first years, when it was packed with mostly traders, heading downtown for the in-person NYC class. Of course, Nassim looked most endearing when asking for the “quants” - and yes, there were 4 or 5. A total group of a hundred students, several returning, several Nassim remembered, who all looked engaged, relaxed, and right at home. This webinar is the modern day version the epically cool professor who has honed his lectures, and no textbook to cover page by page. Its the second coming of that one class from college you still have all your handwritten notes from, which make you actually smile every time you’re organizing your closet.
Day one starts with reassuring the participants that everyone will be fine, even if not a statistician. In fact, the statistics course you may have taken as part of your prior education may have messed you up. The first question Nassim posed to the class is, “what is the average correlation between two unrelated independent variables?”. Us laymen guessed, probably basically zero. But with his screenshare on and his keyboard in math mode, he writes out the equation, runs it a few times, and it returns values as high as 14%. Plenty high for a social scientist to draw conclusions (who runs plenty more experiments that we just did) as statistically significant. The bold statement is made that the R^2 is much social science is actually completely random, intermittent happy coincidences for researchers. But this statement is at the same time not surprising - if you’ve read Nassim’s books. The cold, hard truth is a correlation of 0.4 is closer to nada, nothing, that it is to a 0.5. Nassim then reveals an Economist-style line graph with a slope around 0.5 (it charts reported IQ vs annual income). The data is literally just a bunch of random dots covering the entire quadrant.
How do you make sure this bad science doesn’t happen to you? In other words, if you want to work backwards from a result you just got - how do you decide if it is random or not? The name of this variable is p - representing the chance you would get this result if the two variables were just completely random. More to come on that I guess.
REAL LIFE-isms:
Your risk is a function of what you’re EXPOSED to. In math-ese, say, “F is not F(x)”. An insurer who doesn’t understand a risk does rush off to read books and understand it - they call the best lawyer they have to write a clause into the contract about it. You don’t have to understand the rain in Spain if you have no exposure there. [my question - I’m not in Spain, but how to I know that I don’t have secondary exposure - like goods I might buy or companies I invest in?]
Having no map is better than having a random map, or a bad map. At least when you have no map, you look out the window.
Path dependence matters for a lot of things. You get different results if you wash your clothes and then iron them, then if you iron them and then throw them in the wash. If the plan to get rich requires dying halfway through, its not a very good plan. More practically, if taking a gamble with all your money could mean losing everything at one point, doubling it afterwards won’t mean much - 100x zero is still zero. This concept will lead us nicely into the survivorship bias.