Apologies to those earlier posters who's words I'm just repeating, but ...
If they are not (and it's accepted wisdom that they are not ... i.e. price distributions are reckoned to have "wide tails" ... note earlier poster's alias "Wide Tailz"), then these tail events can happen much more frequently than they would do if the market was a pure, normal distribution.
So, in general market prices don't follow normal distributions ...
Just because market prices don't follow normal distributions doesn't mean they don't follow other distributions. Here's a list of other probability distributions ... http://en.wikipedia.org/wiki/List_of_probability_distributions
A distribution does not have to be a normal distribution for probabilities to be useful.
Your statement ("...should only happen less often than ..." etc) is only correct if price distributions are normal Gaussian distributions.Quote from Took2Summit:
... if the black Monday event that took place in 1987 was an 18 standard deviation move (on the conservative side) which should happen less often then 1 in the entire duration of the universe ...
If they are not (and it's accepted wisdom that they are not ... i.e. price distributions are reckoned to have "wide tails" ... note earlier poster's alias "Wide Tailz"), then these tail events can happen much more frequently than they would do if the market was a pure, normal distribution.
So, in general market prices don't follow normal distributions ...
Quote from Took2Summit:
... How can people claim to have an edge? ... How can you even have statistics on events that all possibilities aren't even known? Statistics on dice or cards are obvious as all possibilities are a given, but how can someone assume they know all possibilities in the stock market? Are edges just delusional?...
Just because market prices don't follow normal distributions doesn't mean they don't follow other distributions. Here's a list of other probability distributions ... http://en.wikipedia.org/wiki/List_of_probability_distributions
A distribution does not have to be a normal distribution for probabilities to be useful.