Last week, thanks to one LinkedIn contact, I found out that Nassim Nicholas Taleb had released a new book. A first draft is available for free here. The new book – named “Probability, Fat Tails and Antifragility: Elements of Risk Engineering” – is a technical development of the issues that the author has been explaining in her last books. I haven’t had enough time to assimilate the whole information yet – likely one needs the entire life to understand it – , but in spite of that, I’d like to stand out some ideas that I share with the author.
Prior to begining, I should admit that I am a Taleb follower. To me, Nassin Taleb is one of the most brilliant minds on finance world. I already wrote here that “The Black Swan” should be a mandatory reading for pure quants. And now, I’d add this new book as a further step. So, the following opinions must be understood under my esteem towards the author.
Asymmetry and Insufficiency of Past Data. This is one of the strongest topics that actually turns out like a torpedo aimed right at the waterline of statistical inference. Taleb suggests that “Real world” becomes much more complex and «fat-tailed» that common theories say. The essence of the empiricism and science is: Look at the world, think about it and try to model the facts you see. However, social behavior and whatever variable that involves the human nature, entails a percentage of uncertainty, big or small depending of each feature, that must be considered. So far, everybody could agree. Nevertheless, Taleb goes further and suggests that this uncertainty overrides statistical inference as well. Why? Because the world is such complex and this complexity is growing so fast than past is increasingly less important. Due to new technologies, the world is more interconnected than ever before. We have a lot of examples in the recent financial crisis. For instance, the Dow Jones Flash Crash of May 6, 2010 started with rumors about budget problems in some European Southern countries. The marketing, the consumer’s taste… Today more than ever, the world is a village. Internet also has revolutionized that. Internet has become economic and financial data much more «fat-tailed».
Errors exist. No mistakes cannot be an assumption. I once heard Taleb say that his life’s mission is try to figure out how to build a society on which people can make mistakes. Frankly, it’s a huge challenge. It seems a self-version of Halon’s Razor “Never attribute to malice that which is adequately explained by stupidity.” So, stupidity or mistakes exist and will continue existing whereas only one human being keep alive. Then, why the vast majority of theories still assume the agent’s rationality? Events like the collapsed of LTC, the rough trader working for Societé Generel or the riot of red flags that nobody saw in the scandal of Madoff, are the bread and butter.
The people’s ability to make mistakes also generates asymmetries. If one thing has happened, we know that can happen again. However, if one thing never has happened, that doesn’t mean that cannot happen ever. We should not be underestimated the mankind’s capacity to get over our bounds through stupidity or even malice.
Focus on the consequences of low probability events more than calculate the exact probability of these events. This topic is a logic consequence of first topic. Historical data can help us to understand the past, but it doesn’t say anything about our current concerns. Past events obviously give us many clues that should be keep in mind, but our current concerns must be thought looking forward, not looking backwards. It means that one analyst must have a depth understanding of his goal, besides technical knowledge. An analyst is a person who understands the world and attempt to avoid the worst scenarios. Perhaps historical data and classical statistical inference can be useful for his target, but is not the essence. That is, an analyst must be, first of all, a cultivated person, and after that, a specialist. And nowadays, unfortunately, it seems to me that the “Uncultivated specialist” is too much spread among line managers and Chairmen of financial industry.
Maybe the main lesson of Taleb’s thinking is that we should not forecast. The world is so complex, so linked, and our forecasts are part of that tangle of relationships, that whatever you say, for sure will never happened. I guess the guys in charge of IMF forecast know what I am saying.
Thanks again Nassim Nicholas Taleb