|
PureBytes Links
Trading Reference Links
|
Some interesting observations on the 4.0 RT quote server
This investigation was initiated to determine what if anything could be done
to preclude or correct tick data overflow (wrap around) which is caused by
the (High - Initial Tick Value) or the (Initial Tick Value - Low) exceeding
32768 divided by the symbol scale factor.
**** CORRECTION OF TICK DATA OVERFLOW
It appears the only way to correct tick data in which overflow has occurred
is to write a program which adds/subtracts multiples of 65536 to the stored
tick data. Someone have a program that does that?
**** PRECLUSION OF TICK DATA OVERFLOW
1. The proper method to prevent tick data overflow would be to change the
scale factor for each symbol in the SU such that the 32768 limit would never
be exceeded during the day. This means that every new issue would require
addition in the SU - a full time task in itself unless some smarty creates a
program that can do all this automatically.
2. A less purist approach is to create Exchanges with different scale
factors using unused regional exchanges (i.e. BOSTON could be scaled to
1/100 and all high dollar volatility stocks moved (Renamed) to that
directory). This works because the quote server apparently does not care
where a symbol resides. In fact, an interesting aspect of the 4.0 quote
server is that a symbol may be duplicated in as many exchanges as desired
and all values will be updated in real time. This allows for an interesting
situation where a high dollar volatility stock such as BRK/B ($2431 at the
moment) could have tick data saved in an exchange scaled to 1/10 (3276.8 max
open to high/low change). It could also be added to an exchange scaled to
1/10000 with no historical data being collected (the standard default) and
this value used for Quote Display. This is a little like 'having your cake
and eating it too' since you can display practically any number of digits on
any value (2431.8125 for instance).
3. The upside to the technique is that the default exchange scaling in the
EU is used eliminating the need to address each symbol in the SU. Plus I
believe I can easily write a program which will automatically move symbols
between exchanges if tick data has not been collected.
4. The downsides to the technique are:
a. visability is lost as the the actual exchange a symbol belongs to and
b. the charting half of the program is more picky about the exchange it
is getting data from and therefore if symbols are frequently moved between
exchanges, the charting symbols/exchanges will have to be re entered.
Although by no means pure, this does seem to be a practical approach to
handling different scale factors.
Looking forward to all comments.
Bob Scott
|