|
PureBytes Links
Trading Reference Links
|
ok here it goes,
1. as most of you use TS, one thing most of you never tried
is to use TS to collect statistics! I no longer use TS
for any system test as it is too limited in so many way...
anyhow, here is something useful,
have you ever tried to "check" if a moving average is
really the statistical mean and the price flutuated around
it are really distributed in a bell curve?
define an array in TS to store cumulative counts of where
the price are relative to a mean (whatever you like as the
mid point) then at the end of the run, send the result to
a printlog.
you can write that as an indicator, just use a dummy plot
of some number to by-pass the checking.
and to send the output to printlog, for TS3/4, if I
remember correctly, you can just test for a specific
date and time which happens to be the last bar, then
just dump the array out.
2. results collected this way tells a lot about the
"reference number" you are using ...
if it is truely normally distributed over a long period of
time (depending on the time frame you use) - then that
calculation alone will make you lots of money because
the calculation you are using have a true mean reverting
behaviour!
this is pure mathematical play,
most people do not even check if the indicators they use
really having the right properties for trading, but just
playing with system twisting ... thus when hitted by
a burst of bad trades based on a system (very common - this
is called exponential arrival of events) - the account got
wipe out ... not even knowing why!
The method I mentioned is just one of the many stand
techniques in time series analysis ... when a "reference number"
process some other properties, or even a "reference method",
then you know you got a statistical edge => then the system
is just an end product of the bias.
time to do some homework - check all the indicators you have
on hand - see which has the right property you can use!
you will be surprised ...
-Lawrence Chan
p.s. I daytrade most of the time, thus high probability is the
name of my game!
----------
> From: gmtac <gmtac@xxxxxx>
>
> ----- Original Message -----
> From: Lawrence Chan <stnahc@xxxxxxxxx>
>
> > > > Mark has given out the model rules already ... if you have
> > coded / traded something like that you would understand.
> >
> Hi Lawrence,
> Mark was very nice to share with us his big experience.
> In his emails there were a lot of true and also some tips .
> But there were no code(obviusly).
> He said us :"I use close of the one hour bar on bond market.
> Counter Trend trading, is the life for me. Computer analysis shows that
> there are many more wiggles in commodity data than there are strait
lines."
> And again:"I know of no other method of trading that makes more money in
a
> shorter period of time if
> done properly.
> For me this was a big surprise!
> My god ,was so difficult learn to go with the trend.
> But maybe I am wrong on what countertrend means.
> You can sell the brekout of yesterday high or sell a breakout of today
high
> but then when and where is the exit?
> You can enter on a dip and then exit on the previous swing high.(this
seem
> better but we can't call it counter trend)
> This kind of strategies are, for me, too much difficult to code so I
have
> never tried them.
> Ok friends, I am ready to learn somethings new.
> Is somebody out there so nice to post one or two counter trend
strategies
> for me!?
> Come on Lowrence share something with us... : )
> Have a nice sunday
> Mario
>
>
|