|
PureBytes Links
Trading Reference Links
|
Elder warned against going too far out in "derivative of derivative"
indicators in his book "Trading for a Living", my main worry about
these kinds of indicators is losing sight of correlation with price.
I think whether double derivative indicators have more lag time
depends on what indicator (normalized, bound, .... etc) and how we
use it. Lag time can be beneficial when it screens out undesirable
noise and synchronize well with other indicators, or it can be very
bad if it makes us buy tops and sell bottoms. Either scenario can
happen, sometimes seemingly at random.
I will use RSI and SRSI to illustrate how derivative indicators can
both be leading and lagging. RSI is similar to Lane's Stochastics
except it does the smoothing before the formula calc giving it an
angular chart pattern which I find easier to read. The formula that
binds RSI also gives it a countertrend bias. At RSI==50 price moves
RSI equally in either direction, the higher RSI is above 50 the more
the same price change will give bigger RSI fall than RSI rise, and
the opposite is true for below 50. The shorter the RSI period the
more pronounced the bias. After a sustained bull or bear run putting
RSI in it's outer limits the very first retrace will move RSI in a
big way, making RSI breakout happen much sooner than price breakout.
Stochastics of RSI (SRSI) exaggerates this countertrend bias as it
moves faster and becomes bound sooner. A system using SRSI crossing
RSI, or SRSI==100 or 0 as benchmark for RSI breakout (SRSI goes to
100 or 0 when RSI makes new highs or lows within the SRSI period)
can signal as fast as the first bar of price retrace.
If the retrace is shallow and price continues up or down the same
timely system can become the worst laggards. The double countertrend
bias has pushed SRSI to the opposite limit of 0 or 100 while RSI
remains a good way above or below 50 on the opposite side of the 50
divide and far from SRSI. A continuation price move will give small
RSI steps and therefore small SRSI steps, small steps and big
distance means SRSI will take a very long time to cross back in line
with price. How long depends on the details of price pattern, a
function of how it zigs and zags and the size of each zig and zag.
Pot luck.
My observations are based on ER2 15 minute bars. Does the same hold
true for other instruments and time frames? Yes and no, a lot
depends on details of price patterns and price patterns when seen as
numbers can have close to infinite variations, not to mention the
large permutation of indicator parameters.
Brian wrote, "On the other hand, I do have a bee in my bonnet about
the absolute rubbish that is put around about trading on the net."
I think it's because anything and everything is both true and false
all depending on context, anyone can get away with uttering
anything. And the industry of Guru Speak is born, thrives and feeds
many people.
"On the other hand, I appreciate that it is *a cheap shot* to make
contrary statements without any supportive argument. At some time in
the future, when I can do a proper job of it, I might lay out the
arguments to support some of the statements I have made about stats
and system design etc. and give the forum a chance to make their
critique."
Please do. BTW my remark about valuing the Pardo book does not mean
I think you made unfair comments about it. Not at all.
"If we standardise to binomial terms 63% wins is very good if we are
considering a coin toss (win value = loss value)."
I think 63% win is phenonmenal, 50-50 coin toss is more the norm,
but what do I know. The straw I grasp is at 35% win I need 1.9 AW/AL
to break even, so I tell myself not to go below AW/AL of 2.
Regards,
Sursod
--- In amibroker@xxxxxxxxxxxxxxx, "brian.z123" <brian.z123@xxx>
wrote:
>
> Continuing.....(I had to go and change a light bulb).
>
>
> My hypothesis that primary indicators (price, vol, OI) are more
> accurate and have less lag than secondary or tertiary indicators
that
> are derived from them is informal and tentative at this stage.
>
> As far as I can tell it is original as I haven't seen any
published
> commentary use those definitions.
> It's probably just a poor man's version of principles that are
well
> known in various academic disciplines.
>
> It's there to be proven wrong.
> Thanks for the two exceptions you provided.
> I have made a note of them and will do a study on them at some
stage.
>
> Re % wins and randomness:
>
> No wins/no losses is relative to %won/%lost.
> If we standardise to binomial terms 63% wins is very good if we
are
> considering a coin toss (win value = loss value).
>
> Brian *:-)
>
> --- In amibroker@xxxxxxxxxxxxxxx, "sursod" <sursod@> wrote:
> >
> > Hi Brian,
> > I enjoy your posts of ideas.
> > I agree with you regarding derivative indicators although I use
> > Stochastics (SRSI) and Bollinger Band of RSI, both derivative of
> > derivative of price. Personally I do not rate any TA indicator
as
> > better than 35-65% accurate and often randomness decides when.
It
> is
> > more important to know the indicator we use so we understand
when
> it
> > might fail then to chase newer and fancier indicators thinking
> > computation sophistication equates the holy grail.
> >
> > I value Pardo's book on systems design very much but ultimately
we
> > have to make our own rules and write our own script, there is no
> > holy bible and no absolute truth except the inconvenient truth
at
> > the bottom line of the trading account.
> > Sursod
> >
> > --- In amibroker@xxxxxxxxxxxxxxx, "brian.z123" <brian.z123@>
> > wrote:
> > >
> > > Hello Bill,
> > >
> > >
> > > I'm pleased you read the reply as I hadn't seen you around for
a
> > > couple of days and thought you might have disappeared.
> > >
> > > Trialling (it's not in my dictionary either - I haven't been
to
> > > school for a while) = = trial and error (paper trading or
virtual
> > > trading where you visualise the outcomes of your systems in
your
> > > minds eye or imagination).
> > >
> > > I base this approach on my theory of success (template) = =
theory
> > > (reading?)followed by practice (backtesting?) followed by
> > application
> > > (trading or virtual trading) and then back to the drawing
board
> to
> > > fix up your mistakes - the path to success is perpetual
> > application
> > > of the above cycle (continuous improvement in business terms).
> > >
> > > I discussed *The Success Template* in more detail in a
previous
> > topic
> > > on Trading Psychology.
> > > It was of minority interest only and some forum members didn't
> > like
> > > it at all so it may not be to your taste either.
> > >
> > > Trading bank = = trading capital = = the amount of money you
have
> > to
> > > trade with.
> > > Depending on your definition of ruin i.e all of your capital
is
> > gone
> > > or 50% of your capital is gone, once you trade to ruin it is
game
> > > over until you can put together another stash (by the way ruin
> > isn't
> > > in my trading goals).
> > >
> > > I understand your point on AccDis.
> > >
> > > I am a liberal and free thinker.
> > > As far as I am concerned feel free to speculate and use
examples
> > etc
> > > in discussions.
> > > There is no pressure on from me to be correct.
> > > It is only a discussion.
> > >
> > > If you are going to write a book etc it is a different matter.
> > >
> > > Brian*:-)
> > >
> > > P.S
> > >
> > > I am not a training guru so feel free not to believe anything
you
> > > can't verify for yourself.
> > > When in doubt put it in the holding pen.
> > > If it doesn't stack up throw it in the rubbish bin.
> > >
> > >
> > > --- In amibroker@xxxxxxxxxxxxxxx, Bill Halliday <halliday_mo@>
> > > wrote:
> > > >
> > > > Brianzee123,
> > > >
> > > > Thank you for your response. I appreciate them very much.
> > > >
> > > >
> > > > Am I on the right track?
> > > >
> > > > Definitely.
> > > > It's a long way to the top though.
> > > >
> > > > Unquestionably true!
> > > >
> > > > Suggest improvements to your approach?
> > > >
> > > > Keep doing what you are doing (reading, thinking,
> questioning,
> > > > discussing, trialling; you are in good company with
AmiBroker).
> > > > It will be interesting to see what you can do in a few years
if
> > you
> > > > are still around.
> > > >
> > > > I will read, think, and question. Discussing my findings
in
> > this
> > > > forum is the way I plan to verify/focus my approach.
> > > > Trialling isn't in my dictionary, what is it?
> > > >
> > > > Suggest improvements to your system development?
> > > >
> > > > I would say to think carefully about the criteria you use
to
> > > evaluate
> > > > your systems (there is discussion in Pardo about this-I will
> > read
> > > it) and to
> > > > understand the principles of money management thoroughly
(there
> > are
> > > > 1000's of systems but you only have one trading bank).
> > > >
> > > > I'm not faminiar with the term: trading bank, what is it?
> > > >
> > > > My initial reason for starting the document was to learn
what
> > a
> > > > system was composed of and how to put one into practice.
> > > > It's a start!
> > > >
> > > > Soon I'll be up to the place where I can set realistic
goals
> > for
> > > a system.
> > > > As you point out, the Accumulation/Distribution System
wasn't
> > > intended
> > > > to be used as a system, but was only used for the purpose
of
> > > learning
> > > > to think/write/plan etc. in "Systems".
> > > >
> > > > Thank you for reviewing my approach and providing comments
on
> > it.
> > > >
> > > > I look forward to your comments on my next offering.
> > > >
> > > > Bill
> > > >
> > > >
> > > > ---------------------------------
> > > > Have a burning question? Go to Yahoo! Answers and get
answers
> > from
> > > real people who know.
> > > >
> > >
> >
>
Content-Description: "AVG certification"
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.5.432 / Virus Database: 268.17.26/670 - Release Date: 2/5/2007 2:04 PM
|