unfortunately my stats abilities fall well short of solving this problem properly (it is over 20 years since I finished my Chemical Engineering degree - with which had almost no stats in it and i have done little or no maths since - I work in the banking industry now!!!!!). I was wondering anybody might have a - beginning to end - worked example of a method to solve my clustering problem. I am trying to develop some analysis for the Foreign Exchange market (in my free time) and need to find clustered price levels of support and resistance (price areas where the price stalls or bounces). I can find an array of prices where historically the price has stalled or bounced but I would very much like to find where these prices are clustered for example say I found the the set of stall/bounce points below:

Oh ... A couple of last points that might be important ... 1) I don't know how many clusters there are so i cannot set it as an input parameter to an algorithm and there 2) there will probably be around 100 values in the array ... with no idea of the the number of clusters and for sure some points will not fall into any cluster at all ... I would define a cluster as being at least 3 points in close proximity

100.01

100.00

100.05

100.10

101.34

104.23

106.43

105.33

110.21

110.22

110.18

I can see visually there are two clusters ... one in the region of (100.00 => 100.05) and one in the area of (110.18 => 110.22), I'd like to computationally find these areas and return the median of the cluster as a likely place the price will stall/bounce ...

Any help would be very much appreciated !!!!!!

Many Thanks

Paul