REFS Online Free Trail
Company REFS: stock picking tool for private and professional investors in the UK stock market
About Jim Slater
How to Use REFS
Need Some Help
Our Data
FAQ's
Contact Us

 

Who Already Uses REFS?


 

d

d

Trend calculation - full technical definition

Equal weight is given to each value in the series, and no adjustments are made to favour recent performance. No distinction is made between a negative statistic with a reducing trend, and a positive statistic with an increasing trend.

The calculation method is described in three stages below. The description assumes the reader is familiar with the following statistical expressions:

helpfile00000307.gif mean

helpfile00000308.gif iteration

helpfile00000309.gif correlation

helpfile00000310.gif weighted average

helpfile00000311.gif compound rate of growth

helpfile00000312.gif discounted rate of growth

helpfile00000313.gif coefficient of correlation

helpfile00000314.gif least squares linear regression

Stage 1: Taking the series of values, the ‘least- squares linear regression’ is calculated, and expressed as a percentage of the mean. The result is translated into a compound equivalent, Rate 1.

Stage 2: A process of iteration is used to establish a notional compound rate, Rate 2, which meets two conditions. First, Rate 2 produces a notional data set which has the same linear regression as the original series of values. Second, if Rate 2 is used to discount the mean of the original series of values to the base date, the result is equal to the base value of the notional data set.

When the actual rate, Rate 1, is in fact a consistent year on year rate of change, the notional data set produced by Rate 2 matches the original series of values, and Rate 2 is identical to Rate 1.

Stage 3: The correlation coefficient of (a) the original series of values, to (b) the notional dataset produced by Rate 2, is used to weight the average of Rates 1 and 2, which in turn gives the Trend result.

If the original series of values correlates perfectly with the notional data set produced by Rate 2, then Rate 2 is the Trend result. If, on the other hand, the original series of values does not correlate at all with the notional data set, then Rate 1 is the Trend result.

If Rate 1 reveals an adverse trend, i.e. a declining positive trend, or an increase in negative trend, the Trend result is taken to be Rate 1. If attempts to establish Rate 2 point to a base value of zero or below, i.e. recovery from a negative position, the Trend result is also taken to be Rate 1.


REFS is available in 3 formats to suit your needs
a a
Updated daily with data direct from the London Stock Exchange
aa
Available monthly or quarterly on CD
a a
Available monthly or quarterly in two hard-copy volumes

 

 © Capital Ideas Financial Publishing Ltd
Sophia House, 76-80 City Road, London, EC1Y 2BJ Registered Number 6445806

Site map / Terms and Conditions