|
PureBytes Links
Trading Reference Links
|
> If you are writing software from scratch, you can use the
>appropriate
> data types for your application -- you have complete control.
>Packing
> data is unnecessary unless you have memory constraints.
Thanks, Steve.
That was great.
Wikipedia/datatypes and the links took me straight to what I needed.
I was thinking about things like defining types, alternative encoding
(implied decimals, fixed point etc), alignment, nibbles, memory
access and all that stuff.
I was emboldedened by the fact that I only have around 3-4 pieces of
data to handle at the lower level.
I was also thinking about words and packing.
I see now that commericial programs are constrained by the common
deniminator of hardware/CPU's etc .... different story to write for
oneself with a purpose bought machine.
Anyway, I found what I was looking for.
I was really talking about the uni dimensionality of memory and how
to correlate that to matrices.
I see now that multidimensional arrays is what I am after (I asked
the forum about multidimensional matrices).
What I am wondering is if a multidimensional array would be better to
store the data like so:
IndexedOneDimensionalArray[A: price,timein,timeout;Bprice,timein,time
out]
OR IODA.price[a,b],IODA.timein[a,b],IODA.price[a,b]
OR as three indexed arrays for price, timein and timeout.
In the end I need the array of arrays:
price[a,b]
timein[a,b]
timeout[a,b]
Since price, time in and timeout are, or could be intergers (same
datatype) they could go in one chronology if it is easier/faster ....
I will have to think about that now.
I don't think C++ is the best array language but I guess it can do a
reasonable job for my middle of the road needs/computer.
Thanks for your help.
BTW my backtester is more of a thought experiment, than a reality, at
this stage.
--- In amibroker@xxxxxxxxxxxxxxx, "hydroblue@xxx" <hydroblue@xxx>
wrote:
>
> Brian,
>
> The PositionScore and PositionSize arrays are an efficient way to
move
> data between the backtester phases. I don't control the interface,
so
> I am doing something unusual to stuff more data through a narrow
pipe.
>
> If you are writing software from scratch, you can use the
appropriate
> data types for your application -- you have complete control.
Packing
> data is unnecessary unless you have memory constraints.
>
> -Steve
>
> --- In amibroker@xxxxxxxxxxxxxxx, "brian_z111" <brian_z111@> wrote:
> >
> > I have been thinking about this, since I started mulling over how
to
> > design a PowerBackTester
> >
> > I will only have to read/write 3 core metrics, in and out of
memory
> > (all other BT metrics can be calc from them on the fly).
> >
> > I have price = 1.0277 (2.77 %)
> > plus in and out time (either datenum or timenum etc depending on
base
> > timeframe).
> >
> > I might also need to store additional data with the trades e.g.
RSI
> > value for ranking signals.
> >
> > For warp speed I am wondering how to get 3 or more units of data
in
> > one memory series .... I wonder if it would be quicker keep all
the
> > data that belongs to one trade together and thenunpack the serial
> > data, in RAM, into 3 the or more arrays required i.e. the
collection
> > of arrays that make up the trade matrix?
> >
> > Don't know anything about bit consumption per integer etc or
> > progamming/computers.
> >
> > I keep coming back to the idea of making 2 * 32 bits operate like
a
> > pseudo 64 bits (and onwards)?
> >
> > Anyone heard of this ... it's probably standard computer practise
or
> > maybe a crazy idea?
> >
> >
> >
> >
> > --- In amibroker@xxxxxxxxxxxxxxx, Dennis Brown <see3d@> wrote:
> > >
> > > How many bits are your numbers?
> > >
> > > Two 11 bit integers can be be packed and unpacked from one 32
bit
> > > float number with ease. You just mask (&) bits to extract the
low
> > > order number. Use multiply or divide to shift the low to the
high
> > > half of the number and back to the low half again. Simple bit
> > > arithmetic.
> > >
> > > Though it seems there should be easier ways to get what you
want.
> > I
> > > am not a CBT user though.
> > >
> > > BR,
> > > Dennis
> > >
> > >
> > > On Feb 7, 2009, at 7:57 PM, hydroblue@ wrote:
> > >
> > > > Thanks Tomasz. I'm looking for a clever way to stuff two
numbers
> > into
> > > > the 32 bit float and then later extract the 2 numbers.
> > > >
> > > > // fraction = 23 bit integer
> > > > // exponent = 8 bit integer
> > > >
> > > > The reason for doing this is to efficiently get an extra
array of
> > > > numbers into the portfolio backtester without thousands of
static
> > vars
> > > > or Foreign symbols.
> > > >
> > > > Is there a way to do this in AFL without writing a DLL?
> > > >
> > > > --- In amibroker@xxxxxxxxxxxxxxx, "Tomasz Janeczko"
<groups@>
> > > > wrote:
> > > >>
> > > >> Hello,
> > > >>
> > > >> IEEE 754 Standard floating point (32 bit, single precision)
> > > >> http://en.wikipedia.org/wiki/IEEE_754-1985
> > > >>
> > > >> Best regards,
> > > >> Tomasz Janeczko
> > > >> amibroker.com
> > > >> ----- Original Message -----
> > > >> From: <hydroblue@>
> > > >> To: <amibroker@xxxxxxxxxxxxxxx>
> > > >> Sent: Saturday, February 07, 2009 5:13 PM
> > > >> Subject: [amibroker] AFL floating point number
representation?
> > > >>
> > > >>
> > > >>> I need to know what range of values and what precision can
be
> > > >>> represented in the AFL floating point arrays, specifically
the
> > > >>> PositionSize array.
> > > >>>
> > > >>> It would also be useful to know how many bits are available
in
> > the
> > > >>> mantissa and exponent. Any help appreciated.
> > > >>>
> > > >>> Thanks,
> > > >>> Steve
> > > >>>
> > > >>>
> > > >>>
> > > >>> ------------------------------------
> > > >>>
> > > >>> **** IMPORTANT ****
> > > >>> This group is for the discussion between users only.
> > > >>> This is *NOT* technical support channel.
> > > >>>
> > > >>> *********************
> > > >>> TO GET TECHNICAL SUPPORT from AmiBroker please send an e-
mail
> > > > directly to
> > > >>> SUPPORT {at} amibroker.com
> > > >>> *********************
> > > >>>
> > > >>> For NEW RELEASE ANNOUNCEMENTS and other news always check
> > DEVLOG:
> > > >>> http://www.amibroker.com/devlog/
> > > >>>
> > > >>> For other support material please check also:
> > > >>> http://www.amibroker.com/support.html
> > > >>>
> > > >>> *********************************
> > > >>> Yahoo! Groups Links
> > > >>>
> > > >>>
> > > >>>
> > > >>
> > > >
> > > >
> > > >
> > > >
> > > > ------------------------------------
> > > >
> > > > **** IMPORTANT ****
> > > > This group is for the discussion between users only.
> > > > This is *NOT* technical support channel.
> > > >
> > > > *********************
> > > > TO GET TECHNICAL SUPPORT from AmiBroker please send an e-
mail
> > > > directly to
> > > > SUPPORT {at} amibroker.com
> > > > *********************
> > > >
> > > > For NEW RELEASE ANNOUNCEMENTS and other news always check
DEVLOG:
> > > > http://www.amibroker.com/devlog/
> > > >
> > > > For other support material please check also:
> > > > http://www.amibroker.com/support.html
> > > >
> > > > *********************************
> > > > Yahoo! Groups Links
> > > >
> > > >
> > > >
> > >
> >
>
------------------------------------
**** IMPORTANT ****
This group is for the discussion between users only.
This is *NOT* technical support channel.
*********************
TO GET TECHNICAL SUPPORT from AmiBroker please send an e-mail directly to
SUPPORT {at} amibroker.com
*********************
For NEW RELEASE ANNOUNCEMENTS and other news always check DEVLOG:
http://www.amibroker.com/devlog/
For other support material please check also:
http://www.amibroker.com/support.html
*********************************
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/amibroker/
<*> Your email settings:
Individual Email | Traditional
<*> To change settings online go to:
http://groups.yahoo.com/group/amibroker/join
(Yahoo! ID required)
<*> To change settings via email:
mailto:amibroker-digest@xxxxxxxxxxxxxxx
mailto:amibroker-fullfeatured@xxxxxxxxxxxxxxx
<*> To unsubscribe from this group, send an email to:
amibroker-unsubscribe@xxxxxxxxxxxxxxx
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
|