I was reading some papers the other day and came across a very interesting paper regarding the 1987 stock market crash. What was particularly interesting was that this paper appeared in Physical Review Letters, which is probably the top physics journal in the world. (I have one paper published in it and have another being reviewed.)
I'll try to summarize the basic points of the paper. Usually stock prices move in small increments that can be modelled by a Gaussian distribution (bell-curve), with large price moves being less probable than small ones (The Black-Scholes model). The authors analyzed a 2-month period surrounding the market crash of 87 and found that there were large fluctuations that were equally probable on many different time-scales in the weeks preceding the crash. Basically it was a breakdown of the Black-Scholes model and made a crash much more probable.
Interestingly, this behavior is quite similar to the behavior of electron spin in a phase transition of a ferromagnetic metal, or in the behavior of heart muscles between heartbeats.
The authors think that their method could help analysts asses risk.
A very interesting paper, indeed. The link is here, but you need to access it through a local library or university.