Today is unique because what happens today isn't exactly like what happened yesterday and what happens tomorrow won't be identical to what happens today. However, there are similarities in what does and does not happen and this data can be recorded and patterns detected.
Suppose you were furnished with this weather history: Over the last 59 years the wind speed averages 9 miles per hour. The month with the highest average wind speed is April, at 10 mph, and August has the lowest average wind speeds of 8 mph. If you recorded the average hourly wind speed for the last 59 years, the wind speed chart would look similar to a bell.
But not necessarily…
Based on this chart, you might say there was a 98 percent chance that winds would be between 1 and 33 mph and plan accordingly. And you'd probably be right, unless tomorrow is August 24, 1992, because this is the chart for Miami, and Hurricane Andrew will hit today with 150 mph winds.
In general the modeling works because tomorrow often is similar to today. The problem is when the assumptions are wrong or when one confuses low probability with no probability. These same probability realities apply to Wall Street, but because human emotions are involved, the unpredictability is much greater than it is when merely forecasting hurricanes.