This is very much a 'work in progress'. In 'Cybernetic Analysis for Stocks and Futures' John Ehlers used the Fischer transform (p.1 - p.10) to convert non-Gaussian stock prices to a nearly Gaussian probability distribution.
I'm playing here. I think I need to calculate Relative Strength (RS) first, then RSI, then apply the Fischer transform. (See:http://blog.traderslibrary.com/sylvain_vervoort/2010/09/smoothed-rsi-inverse-fisher-transform-excerpt-from-sc-magazine.html)