|
PureBytes Links
Trading Reference Links
|
I have 98,898 QCOM ticks on 1/28/99 from 9:30:00 to 16:00:00 from DTN sat.
I use a PII450, 256mb, TE920 serial card at 460.8kbps, TS100 SP4b, NT4 SP6.
I have zero buffer overruns between DTN box and serial port and maybe 300
overruns per day between serial port and GS. The overruns between the
serial port and GS are logged in the file Omega\server\dtn\log.txt which I
monitor and delete weekly since it keeps getting bigger if you let it. I
figure one overrun is at most one quote so loosing 300 ticks per day out of
millions is irrelevant given that I am collecting every tick on 19k stock
and index symbols.
I think the tick counts are at least informative so you know what other
traders are getting. I am interested in trade volume so loosing ticks is
disasterous to me because every tick lost means lost trade vol. I have used
Signal FM and sat, BMI cable and sat, SP Comstock sat and now DTN sat which
gives me the best data and the highest tick counts. My only gripe about DTN
is that you do not get Adv/Dec vol or issues or Tot vol on the exchanges in
separate symbols that can be charted or used in formulas.
Sentinel's numbers confirm that we are getting consistent results from DTN
so I am happy.
As an aside, the number of symbols I collect has no effect on my tick
counts. I restricted my portfolio to a small number as a test and got the
same
tick counts. I immediately went back to collecting every symbol. Makes it
real easy when you want to look at a new stock to have plenty of ticks
stored plus I have absolutely no symbol maintainace to worry about.
Bill Wood
> -----Original Message-----
> From: Sentinel [mailto:rjbiii@xxxxxxxxx]
> Sent: Monday, January 31, 2000 2:45 PM
> To: wayne@xxxxxxxxx
> Cc: Omega List
> Subject: Re[2]:BMI RealTime $59/mth vs TradeStation.com for $???
>
>
> That was on the 28th and included the dribble, between 0930 and
> 1600 EST the
> count I have from DTN is 98,947, for QCOM.
>
> You also have to be careful because some feeds send duplicate
> last prices for a
> bid or ask change, and if you collect Bid Ask in GlobalServer you
> will see more
> ticks than occur when counting ticks on the last price.
>
> I found this to be true of E-Signal while beta testing ProSuite
> on that feed. I
> know that is not the case with DTN. It shows up the most on
> options where bid
> ask may change many times before the last price will.
>
> This is using ProSuite 2000i, a 266 Pentium II, 198 MB of Ram,
> PacificCom Ware
> Turbo 920 Serial Port at 460800 Baud and Windows 95B. I have 1
> or 2 Fifo over
> runs at about 10:30 AM about 3 days out of 5 across my mother board, and 0
> between the DTN Box and SerialPort.
>
>
>
>
> ____________________Reply Separator____________________
> Subject: Re:BMI RealTime $59/mth vs TradeStation.com for $???
> Author: Wayne Mathews
> Date: 01/31/2000 1:55 PM
>
>
> The most substantial results from the recent postings is Bill Wood?s.
> Using DTN (sat) he got 65,781 ticks for QCOM on 1-27. I checked with
> Quote.com and it shows the same number within 20 ticks. My TS100 (sp4b)
> shows 27% less at 51,634 with BMI cable. (Sentinel wrote he got 99,107
> for QCOM on DTN but that must be for 1-28 on which date I got 81,925,
> 21% less ). Comparing with a reference source outside of TS is the only
> way to decide on which setup is more accurate.
>
> I posted the following some time ago (before Dynastore). Most of it is
> still relevant now.
>
>
> Date: Sun, 23 May 1999 09:34:32 -0700
> From: Wayne Mathews <wayne@xxxxxxxxx>
> To: =omega list <omega-list@xxxxxxxxxx>
> Subject: RE: Tick Counts
>
> Well, we are back to tick counting.
>
> The number of ticks you get on your machine depends on many factors. The
> machine speed, ram etc., the speed of the market and how many symbols
> are in you portfolio. Posting your tick count is a waste of time. Your
> TICK COUNT MUST BE COMPARED NOT TO SOMEONE ELSE?S DATA FILTERED
> THROUGH TS, BUT TO SOME RELIABLE REFERENCE SOURCE SUCH AS QUOTE.COM.
> I spent most of last April on where ticks were being lost in TS 4 and
> posted the
> following to this list (how easily posts disappear into the bowels of
> these lists and are forgotten).
>
> > Subject: CL_Best Data Feed, You Already Have It
> > Date: Fri, 16 Apr 1999 18:07:37 -0700
> > From: Wayne Mathews <wayne@xxxxxxxxx>
> > To: code-list <code-list@xxxxxxxxxxxxx>, omega list
> <omega-list@xxxxxxxxxx>
>
> >
> > I have made extensive comparisons of missed ticks between charting
> > in TS 4 and two charting services, Quote.com and BMI. I used BMI
> > cable as the data feed for TS 4. Quote.com is over the Internet. BMI
> > internal (native)
> > charting allows you to see and print Time & Sales (T&S), that
> is, each and
> > ever tick (I believe e-Signal has this feature).
> > Quote.com has the same feature. Since BMI charting and TS can
> > not run at the same time I used Quote.com as the reference. THE NUMBERS
> > ARE PER CENT OF TICKS MISSED as compared to Quote.com. Since it seems
> > more ticks will be missed the faster the market is for that stock I have
> > listed the seconds per trade in parentheses. These numbers are
> > different due to collecting BMI and TS 4 data a few minutes
> apart on the same
> > machine running TS 4 (closing one charting program and opening
> the other).
> >
> > I am listing only part of the results I obtained but the pattern is the
> > same as shown below. (Data is taken with 65 symbols in portfolio and
> > 166MHz, 64M, 2.1G HD. My second machine is 400MHz, 128M, 6.2G HD
> > on which I have TS2000 loaded and the results with TS2000 are similar)
> >
> > Stock BMI (sec/trade) TS (sec/trade)
> >
> > AAPL 1.7% (10.1) 10.3% (17.5)
> > AOL 3.9% (3.2) 30% (1.9)
> >
> > (There were a string of losses for 20 seconds for AOL
> on BMI and if it
> > were not for those the % would be 1%. I look for consistent patterns.).
> >
> > What the results tell me is that BMI is not the culprit for lost ticks.
> > IT IS THE TS 4 SERVER!! REMEMBER, ALL THE TICK RESULTS POSTED ON THIS
> > LIST ARE READ ON (OR THOUGH) TS. IF THE TS SERVER IS AT FAULT THEN
> > COMPARING SATELLITE FEED WITH CABLE WITH FM IS NOT GETTING AT THE
> > PROBLEM SINCE ALL THOSE GO THROUGH THE TS SERVER! I really can not
> > believe it is that hard for paid, professional programmers to get this
> > right. And why does not Omega advise us of these type limitations? Let
> > me make it clear -BMI IS AS GOOD A DATA FEED AS WE ARE GOING TO GET FOR
> > TS 4.
> >
> > Wayne Mathews
> >
>
>
>
|