Re: Intra-bar Signals, hypothetical high resolution platform.

["Bilo Selhi" <biloselhi@xxxxxxxxxxx>]

here is a hypothetical example of how should
high resolution data be used with low resolution data
in a futuristic trading platform:

for example, return  me volume of first 10 min of 30 min bar data 3 bars ago.
so, high resolution = 1 min and low resolution = 30 min
we want to count 3 30min bars back, ie third bar back from the current
bar and we want to get the volume of first 10 min of that third bar with
1 min resolution.

return = volume[3](1,1-10);
read as: count 3 bars back on 30 min frame, take 1 min resolution and get me
volume over 1st 1 min bar to 10th 1 min bar...
the machine will count 3 30 min periods back, then accumulate
volume over 10 minutes starting with the 1st minute and return
it to you.

another example would be to get the highest high instead of volume
so the function call would look like this
return = highest ( high, 3 )(1,1-10);

the first part highest ( high, 3 ) is no different from current TS function
the second part gives reference to higher resolution data within that bar:
(1,1-10)  ie 1 min data, from 1 - 10th bar of that data.

here is a call to get the range between 10 and 15th minute within a 30 min bar 5 bars
back.
return = prange[5](1,10_15);
sez, count 5 30min bars back, give me range of price movement from 10th minute to 15
minute
on that bar...

here is a buy signal to buy exactly 5 minutes after the start of the new 30 min bar

buy(1,5) market;   sez on the fifth minute of the current bar using 1 min resolution.

just imagine things you can do if you have access to data structures like that!
and this is just hypothetical, the data structure representations  can be polished  up to
allow for the most flexible access to data streams.

then you should be able to set the "update" rate on 30 min bar all the way to
1 second or a tick if you wanted to...
the backtest resolution will be based on the highest resolution you access in
your system...
system accuracy will be maximized and closest to the theoretical limits.
modeling accuracy will be the highest.
bilo.
ps. when structures like that will be available the majority of the academic
research on discrete financial data will be OBSOLETE... :-)
this will require a lot of broadband, faster machines and the new trading
platform. could be done though even now...