RSI: RELATIVE STRENGTH INDEX. A REVIEW.

I have never been a partisan of TECHNICAL ANALYSIS because of its lack of accurate, orthodox and non-trivial theory regarding accrued and expected prices on assets and derivatives (e.g. Candlesticks, Japanese Candlesticks, head and shoulders and so on)

Nevertheless, I have to save the best of it. A mathematical ratio named: RELATIVE STRENGTH INDEX.

This ratio help us to measure how many percentage points of positive changes on prices or quotes are accrued per percentage points of negative changes on prices or quotes; given the same asset or derivative under surveillance.

It is an excellent measure of the dominion held by bidders and askers and, in the long run, shows us, what is the best position to open when trading: wether long or short

Unfortunately, probabilities are not being taken into account leading this tool to misleading and deceitful performance. The classical formula states:

p* / q* = RSI.

Where:

p* = average percentage points after positive changes on prices on a given security

q*=average percentage points after negative changes on prices on a given security

The flaw and failure lays in the fact that both averages, according to this formula, have equal weights or chances to happen. This is thoroughly against basic PROBABILITY THEORY

Assuming yields behave continously compounded and data sets are discrete, an orthodox proxy would look like this:

        

p* (np/N)

__________ -1= RSI

q*(nq/N)



p* = average percentage points after positive changes on prices on a given security

q*=average percentage points after negative changes on prices on a given security

np: number of data from positive changes of prices

nq: number of data from negative changes of prices

N: number of total data points or (np+nq)

You may calculate better with an inverse formula, so the absolute negative strenght will be on top, and you might find out savvy, witty and clever investment have indexes between -1 and 0

q*(nq/N)  / p*(np/N)  = RSI

A brief example: an stock has a positive average on its price equal to 1% daily, the same stock has a negative average on its price equal to -0.6% daily. We have got a 100 data sample and the probability of the price increasing or break even is 75% while the probability of decreasing is 25%

(-0.006*0.25) /(0.0099* 0.75)  = -0.20202020. So, in this situation, a long position serves well the interest of the investor.

Thanks a lot to Dr. Samuel John Gilliland for "healing the wounds and injuries" on this note