Great piece from the oddhead blog here. Great - but in my analysis - wrong....
The markets are not in any shape or form random, in as much as they are not completely unpredictable.
It is to be expected that an analsyis of any daily data would show a random distribution. Movements over relatively long time periods have a stronger, self fulfilling, prohetic capacity due to the effects of mass psychology. Many traders can hold a contrarian point of view for an hour. Fewer, one that contradicts the concensus for a day. Very, very few individuals can hold out for a week, month or year.
In other words, with enough time to come to a concensus, the market tends (not all the time, but tends) to revert to an "established view". The majority of people thinking "the market has dropped 2%, it'll probably bounce back to a fib. point @ 33%" - and on avergae - most of the time, it happens. The net effect being a consistency you see in the oddhead data. where you see an even distribution of larger and smaller price movements to reflect the different states of markets over time.
If you did this analysis with the week or month stats - it would be even more perfectly distributed.
It does not mean that the markets are perfectly random in the size of their movements however. Not at all.
To find human characteristics. To uncover an edge, you need to look in a little closer. If you were to compare histograms on the hour data - 8:00am compared to 10:00pm for example - you'd see extraordinary differences. At 11 o'clock the histogram would be narrow and taller than at 10:00pm. Price movements would be, on average larger, and more frequent.
Its this short term volatility where a lot of traders find an edge.