[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: RSI Formula



PureBytes Links

Trading Reference Links

Thank you for your very clear explanation of measuring 
characteristics of RSI. 

Have you taken a look at Cardwell's concept of price target 
prediction, using what he calls positive or negative reversals?

If price in an uptrend RSI (wilder)14 will tend toward 40 - 80 range:

RSI new low pivots, unconfirmed by price, signal positive reversal:

A measurement of current price low, minus previous lowest price* 
is added to most recent price high, to project next price target.

M.R.

* This lowest price bar will be located as far back in time, as the 
RSI newest low point exceeds all previous RSI lows. 

I think this is approximately a correct explanation.




-- In amibroker@xxxx, "Dimitris Tsokakis" <TSOKAKIS@xxxx> wrote:
> In some recent mails I read some things about division by zero,
> uncertainities, loss of sensitivity etc., referring to the RSI 
formula.
> Let us clear the subject to avoid misunderstandings.
> In Amibroker 3.70 User?s Guide we have
> 
> RSI=100-[100/(1+U/D)] 
> U = average of upward price closes (EMA of gains) 
> D = average of downward price closes (EMA of losses)
> 
> The formula 100-[100/(1+U/D)] is equivalent to
> 100*(U/(U+D))
> which will never suffer from division by zero and any uncertainity.
> Above formula is, like any well defined formula, equally sensitive 
> in its definition domain.
> The only slight difference is that in built-in RSI, Wilders average
> is used instead of EMA stated above.
> As for the U and D, it is not exactly an average of price closes but
> an average of price differences C-Ref(C,-1).
> For a better AFL understanding, you may use the code
> 
> /*RSI 12*/
> t=12;
> Up=IIf(C>Ref(C,-1),abs(C-Ref(C,-1)),0);
> Dn=IIf(C<Ref(C,-1),abs(C-Ref(C,-1)),0);
> Ut=Wilders(Up,t);
> Dt=Wilders(Dn,t);
> RSIt=100*(Ut/(Ut+Dt));
> Graph0=RSIt;
> 
> Above code will give the graph of built-in RSI(12) function.
> 
> Dimitris Tsokakis
> PS: It would be interesting to extend this in some future edition
> and have
> RSI(Array,Periods=14)
> and the same for 
> StochK(Array,Periods=14) and StochD(Array,Periods=14)
> This will give the possibility of easy reference to forms like
> RSI(High,12) or StochD(Low,20) etc., interesting for Research.
> RSI and Stochastic are interesting Mathematical tools and may
> be applied to any Array.