Mandelbrot, a mathematician converted into an economics professor, describes some shortcomings of the classical understanding of financal markets. His anaylsis of a century's stock market data shows that especially two prevailing assumptions - gauss distribution of price changes - non-correlation of events are not at all true. Taking some similies from the Bible he calls them - Noah effect (existence of extremely large outliers; e.g.rainfall) - Joseph effect (7 good years followed by 7 bad ones)
The book contains some explanations about fractals and a lot of history. Most chapters are humorously worded and real fun to read.
However, while conceptually very original, at some point the detailed description of his new model gets a little diffuse. Maybe this is part of the fractal legacy: interesting but by nature chaotic.