|
PureBytes Links
Trading Reference Links
|
I have been doing some qualitative testing of the BMI satellite datafeed
with both the Omega Version 4 Server and the UMDS Version 2.0 plus Quote.com
data using cable modem internet delivery. What I have seen so far using
INTC as a reference is absolutely terrible. I think many of us realize that
the 38 Kbaud BMI port probably can't handle all of the peak trading data but
I never would have guessed the results which I obtained. If indeed 38 Kbaud
is too slow, then BMI is probably performing a balance act between data
delay and dropping ticks but my preliminary tests show the situation to be
grave (as in dead).
The initial test was to determine if a serial buffer was required with a new
dual PIII PC and I decided to use the UMDS as a reference for Omega Data
Server performance without the buffer and also observe Quote.com data just
for kicks. I computed tick rate at 5 to 15 minute intervals throughout the
day for INTC for the Omega data server, the UMDS and Quote.com. The results
of this data indicate that during the first half hour after market open the
observed BMI tick rate was reduced from approximately 90% of the Quote.com
tick rate to almost 40% Quote.com tick rate (that's a 60% data loss). From
this point the relative tick rate slowly increased to 100% over the next 2.5
hours. The 100% level was then maintained until approximately half an hour
before market close when it gradually reduced to approximately 65% of the
observed Quote.com rate.
Bottom line - if up to 60% data loss for almost half the trading session
bothers you then maybe we should try to get something done about it. I have
both BMI satellite and Quote.com and would not be without either but I trade
primarily on Quote.com data (if it's running). BMI data seems to lag by
about 30 seconds at market open and typically 5 or 6 seconds during the rest
of the day. I would switch to Quote.com in an instant but currently my
experience is that their internet delivery is not that reliable.
Although I have used only one symbol for my test purposes, I suspect that
all data is "hit" in a like manner.
The BMI (DBC) folk may not listen to us individually but if a block of users
were united we might have better luck. If you are interested in continued
investigation please voice your interest. How many BMI users (satellite and
cable) do we have in the audience?
In order to not interrupt "the list" too much please respond either by
e-mail directly to me or add comments to my web page Guest Book at:
http://members.home.net/rlscott/guestbook.html
If you don't mind, you could include your name and BMI User number.
Bob
|