|
PureBytes Links
Trading Reference Links
|
SLwiser,
When H=50, L=48 and C=49.9, it is obvious that there was a buying
pressure all day long.
If H=50, L=48 and C=48.1, then sellers were pushing prices all day
long lower and lower.
This is a daily fact not depended from the first sight to absolute
prices.
Of course the measure will be related to absolute prices.
In the examples above, Dv=H+L-2*C would be -1.8/+1.8 respectively.
Two years later an H=5.00, L=4.80, C=4.99 [or respectively
H=5.00, L=4.80, C=4.81] would give Dv -0.18/+0.18.
Since Dv is an absolute measure, the exponential damping gives a
crude description of the price deflation .
For a relative comparison, a quantity like (H+L-2*C)/C or (H+L-2*C)/
(H+L) would be better.
But, the interesting point is that this Dv follows the rate of price
devaluation. This means that the market changed prices but did not
change habits : In a crazy "buy" day, like May 8, they pushed CSCO to
close at 16.27 with H=16.3. The same sentiment on Apr 17, 2000 pushed
CSCO close at 66.5 with H=66.6 for this day [some people bought CSCO
the last quarte of this Apr 17 to...catch the trend.
I have some plans for exponential curves instead of steady/parallel
lines. Some quantities, like stochastics, are pure oscillators, they
have a [0,100] range independent of absolute prices.
Some other quantities, like Dv=H+L-2*C or D-ratio=(H-L)/(H+L) have
oscillator properties but have also some relation with absolute
prices [they are quasi-oscillators]
To handle a quasi-oscillator with paralle steady levels is not the
best idea for building up trading rules.
We have examples of D-ratio systems with lack of signals the last one
year or so.
An exponential curve system, like the first attempt here, would give
better description, instead of a [14,43] crossover.
Quite advanced topic perhaps but, if the market is exponential, then
we should use exponential tools and, perhaps, you agree.
I will try to materialize this theoretical thoughts and come back
with details.
In general, I prefer to "read" the market, instead of "force" the
market to "fit" to some old-fashion models. The reason is quite
simple : Nobody can "force" the market follow any model. The sure
result will be to loose your money.
Reading the market without prejudice may give you some sweet profits.
In other words, if the market is linear, trade with linear logic, if
it is exponential, do not hesitate to step in.
The usual human argument saying "linear is scrutable, exponential is
complicated" is not always correct...
Do not forget that RSI or ADX are quite complicated mathematical
forms, like the majority of T/A tools.
Thank you very much for your comments.
Dimitris Tsokakis
--- In amibroker@xxxx, "slwiser" <slwiser@xxxx> wrote:
> Dimitris
>
> I usually do not say much but looking at these three graphs, it
> appears to me (course I could be wrong) that the expoential aspect
of
> these graphs is a result not of buying/selling pressure but the
> actual value of the stock over these periods. MSTF has maintain a
> value between 50 and 70 over this period, while Yahoo has dropped
> from well over 100 to around 10 during this period. QQQ has
dropped
> from over 100 to around 25 during this period. The range between
> these are 1.4 for MSTF, over 5 for Yahoo and over 4 for QQQ. This
> value ranges should be normalize to make it accurate I would
think.
> Maybe the ATR provide a better picture than the H+L-2C that you use
> to show buying pressure. Maybe just normalize the stock values
> before doing something like your doing would work.
>
> Just a thought. BTW, keep up the good work. Ideals are what we
need
> to work with and you are an ideal person.
>
> SLwiser
>
> --- In amibroker@xxxx, "Dimitris Tsokakis" <TSOKAKIS@xxxx> wrote:
> > Close value is not always the average of H, L. If there is a
strong
> buying pressure, Close will be near H and
> > if the selling pressure is strong, Close will be near L.
> > The quantity Dv=H+L-2*C is descriptive enough.
> > For many stocks the Dv graph follows the last 30 months an
> exponential damping, as you may see from
> >
> > Dv=H+L-2*C;
> > Plot(Dv,"",9,2);
> > DvH=LastValue(Highest(Dv));
> > DvL=LastValue(Lowest(Dv));
> > Coeff=0.005;
> > A1=DvH*exp(-Coeff*Cum(1));
> > A2=DvL*exp(-Coeff*Cum(1));
> > Plot(A1,"",1,8);Plot(A2,"",1,8);
> >
> > Exponential coefficient 0.005 is satisfactory for many examples.
> > In some cases ( YHOO) you may go up to 0.0075 or come down to
> 0.0025 (MSFT).
> > Some stocks do not follow this exponential decay model.
> > Dimitris Tsokakis
|