Over the past few years, one of my back-burner projects is learning how to make an automated financial security trader. My trader/economist friend first proposed the idea to me 4 years ago, so I started educating myself about this ever since. I was aware of the demise of
Long Term Capital Management, a hedge fund with Nobel laureates on its board, so my gut feeling was that we don't know much about the nature of risks and that the first priority in any automated trading should be: don't get wiped out. I read quite a few papers but I suspect that the authors might be the "often wrong, never in doubt" types (or at least "often wrong, often in doubt but I won't tell you" type.)
By chance, I stumbled upon two excellent books by
Nassim Nicholas Taleb called
Fooled by Randomness and
The Black Swan. These books argue that human often impose patterns on to something essentially random and we don't have a good sense about the impact of rare events. You can see the fall of LTCM and the current financial meltdown as manifestations of the ideas in these books. I will make sure that my children become exposed to these ideas when they learn about probability and randomness.
--- - ---- - ----- ---------
P.S.
5. There are fields, such as physics, biology, and chemistry, where experts are significantly better than non-expert at predicting how reality behaves. Then there are fields, such as financial forecasting, government, and alternative medicines, where "experts" don't know much, if at all, more than non-experts. So, a lot of times, when you think an expert is bullshitting, you are probably right.
6. I notice, while teaching my students about
Bayesian inference, that it's hard or impossible to get true probability distributions for most interesting questions.
