Taleb makes some valid points but it's a little ridiculous to blithely assume that everyone always underestimates the probability of rare events and thus all mathematical finance is useless. There's a whole field of study -- behavioral finance -- dedicated to discovering and understanding the facets of human nature that cause us to over and underestimate probabilities. Financial models often make simplifying assumptions that can make users complacent -- e.g. log-normal distribution of prices -- but they can also lead to a good number of useful facts and methods if you're aware of the dangers of the assumptions.
I think Taleb is taking the easy way out. Rather than doing the hard work of modeling fat-tailed probability distributions, or studying how people mismanage risk, or building new financial models with more realistic assumptions, he throws up his hands and says that the future cannot be modeled. To his credit, he did once run a hedge fund based on his principles (called Empirica), but it was forced to close. Whether this was due to a lack of rare events or to mismanagement, no one can really know.
>Rather than doing the hard work of modeling fat-tailed probability distributions.... or building new financial models with more realistic assumptions...
Unfortunately, you are wrong and Taleb is right.
The future of a dynamical system such as market cannot be predicted any better than a weather can be predicted one month from now.
You could try to 'model' the temporal structure of the market (trying to find and describe its 'strange attractor'), but this is pointless because this market system has too many dimensions and your measurements of the system are too noisy and incomplete.
The best you can do is to run some 'local' predictor. Like simply build a black box model based on the system's past and predict tomorrow based on what you've seen in the past. This works quite well but only for the very short term horizon. And it is not the 'model' you are thinking about.
You should dispose of the traditional statistics altogether. It simply does not work in real life and all of the kludges developed to save it (like ARMA models etc) are pointless.
As far as his fund is concerned, I would think he has done right by stopping trading in times when his strategy did not work. His strategy was to buy out of money options. Well, we had almost ten years of falling volatility and he would bleed to death in such an environment. A log of smart traders also went out of the market (Soros etc), some are coming back now.
Taleb is not even such an innovative thinker. All of the stuff he is talking about has been an active field of scientific study, but he is valuable for being a popularizer of these ideas.
I'm not saying that you should try to predict the future of the market. I would agree with you that in many cases, this is a fool's errand.
Mathematical finance is about a lot more than prediction, however. Risk management is even more important, since there are so many market participants who find themselves exposed to risks that they must hedge. Even your neighborhood mortgage banker is exposed to interest rate risk -- both the level of rates and the volatility of rates. Any multinational corporation is exposed to foreign exchange risk. An option trading desk will often find itself exposed to volatility and correlation risks. It would be foolish to leave these unhedged, and without some sort of a model you're forced to. The right way to approach this isn't to treat markets as some inscrutable thing, but to propose the best method you can and be aware of the risks you remain exposed to.
I don't think Taleb disagrees with this part. What he found annoyed is the modelers frequently forget what we have is a model! While in physics, any elegant theory or model will be "disapproved" by experiments. Financial modelers may just stubbornly refuse to consider the model is wrong! Unfortunately whether string theory is right or wrong may not affect out daily life. But a blow up in millions people's retirement money is in fact a tangible disaster that affect many people's life.
What I get after reading NNT's books is "being humble, prepare to be humiliated" I think it also the same like what PG said for startups, "get users feedback asap, and iterate it fast" An arrogant developer or hedge fund manager is just a disaster waiting to happen.
The only true hedging is - you got an asset - you sell it forward. You are hedged.
What passes for hedging and 'risk management' today in the financial industry is something different. You measure how one instrument has correlated with another instrument (or a basket) in the past build an offsetting portfolio based on these correlation matrices -- and voila!
But your assumptions that those correlations between assets will stay the same in the future (including volatility etc) are baseless. As you see, it is also predicting the future. And it is wrong.
It's often impossible to sell an asset forward, especially an illiquid one. Suppose you are a trading desk that makes markets in corporate bonds. If a client comes to you and wants to sell a given company's bonds, chances are that you will not be able to immediately find a buyer. You will probably be forced to hold on to the bonds for a bit. Now you're exposed to interest rate and credit risks. What do you do? You can hedge some of the interest rate risk by paying on interest rate swaps, and you can hedge some of the credit risk by buying protection in the credit default swap market. But without some kind of model, it's going to be impossible to determine how to use these tools to hedge your risks. Your only other options are to guess, or to do nothing. If you are a large player in the markets, this is extremely dangerous.
Certainly, it's foolish to assume that you are ever perfectly hedged. In this example, several things could go wrong: e.g. credit default swap traders could have a different view of a company's creditworthiness than corporate bond traders, forcing you to pay exorbitant prices for protection on the bonds. But this is a smaller risk than holding the bonds outright.
In short: models can be helpful in risk management, as long as you are aware of the residual risks you are exposed to. Just don't allow them to make you complacent or overconfident.
I remembered a WSJ interview of NNT last year about Empirica after subprime blowout. According to Taleb, Empirica is not forced to closed by lacking of rare events and they did not lose their raised fund, but he felt bored. Because Empirica built its position in taking small loses everyday, and Taleb said that is really frustrating for people operate it psychologically. Just like what Taleb said (not verbatim): human being needs once in a while endorphin kicks in brain to live. And knowing your are losing money everyday to wait the payoff is really not that exciting. Don't you think it is the same feeling when you run a start up, living cheaply and waiting for the pay day? Unless a man with superhuman will power, it is too hard for most human being.
But after subprime, he said he was interested in coming back to market because he thought it would have fun because the volatility of market may come back. Since he has "fuck off" money now, for him the fun is more important.
Is it better than "Fooled By Randomness"? That one was an enjoyable read, but it wasn't really a well structured book. It seemed more like "random" ramblings, all with essential the same theme. Still enjoyable, as I said, but I expect Black Swan to be just the same, so I did pass on it so far.
Black Swan is written in the same style and has the same content to rambling ratio. Same points made, too. Not required reading. Makes a good "I have a 2 hour flight" book.
I listened to Taleb speak at the Long Now and all he really said was, "The future is hard to predict." He was either underestimating the crowd and pulling his punches, or he had very little to say in the first place. Either way, I'm passing on The Black Swan.
All his point in fact is just "Instead of trying to predict future, how about let's just prepare for worst first?" So instead of computing probability of stock market index rising or falling in june 2009. The strategy is just "if all our current holdings dropped 75% in values? what is our protection against such happening". Instead of trying to bet on probablity, he just tries to hedge when black swan happens.
A very nice interview. I found the long lost intellectual delight when I read NNT's "fooled by randomness" and "black swan" last year.
Two writers that I admire most lately are PG and Taleb. I feel amused when their writings sometime just create a storm because most readers feel so upset from what they write.
One most interesting part from this interview is about Taleb's fitness program. PG's "You weren't meant to have a boss" just popped out from my mind. It looks like both of them though walking down different paths in life have found the same reality that rest of the world seems unaware of or long forgotten.
I'm reading his book now, and it's excellent. And I rarely enjoy non-fiction 'meme of the day' type books.
His book is like conversation with an older, intelligent person who has the kind of wisdom you get from life experience. It's also very relevant to anyone working in a startup.
starts getting interesting on the second page, worth a read for the fantastic quote:
“Scientists don’t know what they are talking about when they talk about religion. Religion has nothing to do with belief, and I don’t believe it has any negative impact on people’s lives outside of intolerance. Why do I go to church? It’s like asking, why did you marry that woman? You make up reasons, but it’s probably just smell. I love the smell of candles. It’s an aesthetic thing.”
I think Taleb is taking the easy way out. Rather than doing the hard work of modeling fat-tailed probability distributions, or studying how people mismanage risk, or building new financial models with more realistic assumptions, he throws up his hands and says that the future cannot be modeled. To his credit, he did once run a hedge fund based on his principles (called Empirica), but it was forced to close. Whether this was due to a lack of rare events or to mismanagement, no one can really know.