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

Re: New version of ROC?



PureBytes Links

Trading Reference Links

At 02:44 PM 4/29/98 -0700, Kovacs, Ross R wrote:
>Just wondering if anyone has seen this idea before:
>
>Rate of Change and Momentum indicators have been around for years.  A 
>significant problem has been their often bumpy movement due to changes 
>in price N days ago rather than the change in price from yesterday to 
>today.  In other words, Momentum is
>
>PRICE today  -  PRICE N days ago
>
>As the indicator moves forward in time, Momentum can make big moves 
>that are caused by a new value for PRICE N days ago.
>
>Most technical analysts know that a simple moving average has a time 
>lag approximately equal to ½ its time length, i.e., a 12 day SMA has 
>approx. 6 days time lag.  Doesn't it make sense to use the value for a 
>simple moving average N/2 days ago in the Momentum formula, rather 
>than the single price n days ago?
>
>PRICE today  -  N day SMA from n/2 days ago
>
>should partially alleviate the bumpy movement in ROC or Momentum.
>
>Obviously, this isn't a big idea.  It's just that I've never seen it. 
> Has anyone seen or tried this version?
>Just wondering.
>

I just don't believe one moving average can do it.  If I were to choose, I
would want at least two exponential moving averages, one somewhat short-term
and another somewhat longer.  Once could be 9 days and the other one 30 days
all the way up to 200.

I would also want to back-test them with different alphas.  A sedate alpha
of 0.2 might do, or a more nervous one (favoring the most recent samples) of
0.3 or more.

Were I not forced to give up anything, I would also want a stochastic 
indicator, an MACD, etc.

No single one will do.

PeteNa9090
petena9090@xxxxxxxxxxxxxx