I've been looking at some day session data on the NASDAQ 100 in order to gauge the frequency and range of points moves.
Using data from 1st May 2009 to 27th May 2011 I looked at daily points ranges (for the day session), and the frequency of different ranges during this time period, and charted them.
This shows that there are very few days where the range is 10 points or less, and that there are very few high range days, which is what you would expect to see.
The most common points range is 20 points. 20 point range days occurred on 31 separate trading days for the time period charted.
It appears from the chart that the bulk of points ranges for the day session appear to be between 14 and 29 points.
Looking at it from the perspective of a bell curve, the most common points ranges are between 12 and 31, these occur 68% of the time. That is one standard deviation using the data shown. Day sessions with ranges above or below that are therefore less common, found only 32% of the time.
Of course there are many variables here such as the selection of data used. I used from 1st May 2009 to remove the volatility from the tail end of the crash, that bottomed in March 2009 (the volatility would skew the data). Periods like that are exceptional. Another factor is that clearly as markets rise in value their day session range would tend to extend also, and as some of the dataset was from when the NASDAQ 100 was several hundred points lower, this is likely to have skewed the end result to a degree. This is likely to mean that the 0 to 1 standard deviation range of 12 to 31 may actually be around 14 - 33, based on the present valuation.
Volatility is also something that needs to be examined, and the impact that has on day session ranges. That is something else that can be looked at, at some point.
For the trades I take I also keep stats afterwards of what could have been achieved over the whole move without using a trailing stop, to see the maximum points available from each trade. This is so that over time I can gauge what a reasonable expectation of points might be, for the trades that do particularly well. Looking at the data from the trades I initiated, there were only 4 trades with a possible yield of over 30 points (and that's nearly 100 trades). So that's certainly in line with what I've shown here.
There is an argument to be made for a different approach to stop-loss management here, and that is, if a trades get above a certain number of points, eg towards the 31 figure, the stop should be moved closer tp the price, since at that time the case for a larger move is diminished. I may also look at a more dynamic approach such as reducing my trailing stop as a trade gets over 20 points, and gradually reducing it further as it gains more, to keep the trade in line with the bell curve distribution of day session points ranges.
This is with a nod to Niederhoffer - Quantify Quantify Quanfity