Announcement

Collapse
No announcement yet.

Esignal- Index question

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Esignal- Index question

    I am a new customer who is in the process of evaluating data quality. I have so far noted that ticks are universally missing for the cash indices and therefore I assume all financial instruments. I use cash index data for my initial evaluation, because the exchange has a controlled timed release. An example is $SPX which transmits at 15 second intervals. A list as provided by eSignal is http://www.esignalcentral.com/suppor...esigind_ac.asp

    Does eSignal consider this an issue or for it to be acceptable? If it is something they consider wrong; How long to fix?

    Thanks
    Last edited by BIG1RED9; 05-08-2005, 03:36 PM.

  • #2
    Hi BigRed,

    There's a huge difference between "missing ticks" and update frequency. Most indicies are, by nature, snapshots of the market and hence have some update frequency assigned to them. We carry well over 1000 indicies so there's no single rule that governs all of them.

    Some we calculate while others are sent to us calculated and in the frequency chosen by the source. Even if we do our own calculation, many times the source dictates how the indicie should be programmed and how often it should be sent.

    You may find some vendors that update more often on a particular index and others that update less often. Certainly up to you to determine your trading needs.

    We have adjusted some of our indicies to update more often over the years so if you'd like to make a suggestion on any specific index, feel free to email [email protected] (where all product related suggestions shoudl be sent). From a calculation standpoint, it's certainly quicker to determine the value of an Index like the Dow, with it's 30 components, vs calculating the Russell 3000, with it's 3000 components.

    In the meantime, you can refer to our Indicie look-up page to see the frequency rate for most of the major indicies.

    Thanks.

    Comment


    • #3
      I have reviewed your index information page and am using it as the basis by which to measure. If you lookup time and sales data for $SPX you will note that not all ticks update every 15 seconds as per your information sheet, the CBOE and a comparative data source I am using. In some case there is a loss of more than 20% of the data per day. To me quite serious and it changes my mathematical study.

      Don

      Comment


      • #4
        Don
        FYI if you want to show images in your post you either need to attach them (using the Attach File button in the page in which you compose the message) or place them inline by saving them first to FileShare (or other web page) and then inserting the URL enclosed by [ i m g ] and [/ i m g ] (all without spaces) in the body of the message
        Alex

        Comment


        • #5
          Attached is time & sales for $SPX for easy reference.


          Last edited by BIG1RED9; 05-09-2005, 06:43 AM.

          Comment


          • #6
            Hi,

            In your example, $SPX has 4 updates per minute and are consistently 15 or 16 sceconds apart (rounding for time will cause the 1 second display variance since we don't display milliseconds).

            I should also add that if an index's value does not change, we don't re-send it. That will account for gaps that appear to be 30 or 45 seconds in some cases.

            Please clarify further what you mean by a 20% variance?

            Thanks.

            Comment


            • #7
              Based on 15 second intervals, there should be 1560 ticks per day.

              I have included information on $SPX, $GSO and $GSM all of which generate 15 second ticks and some of the indices I trade from.

              Date - $SPX $GSO $GSM
              5/2 -- 1415 1139 1222 (Ticks)

              5/3 -- 1434 1550 1570

              5/4 -- 1460 1529 1620

              5/5 -- 1403 1167 1266

              5/6 -- 1443 1504 1622
              Last edited by BIG1RED9; 05-09-2005, 08:28 AM.

              Comment


              • #8
                1560 is the approximate max. # of updates but you can see a handful of pre and post market updates as well. Also, as mentioned previously, if the value of the index is unchanged from the previous update, we won't re-transmit it. Therefore, its unlikely you'll ever see exactly 1560 ticks on any given day.

                I know of no easy way to determine how many ticks "should" have been sent but I can tell you from scanning T&S, I do not see any irregularities. If you spot one, feel free to re-post.

                Thanks.

                Comment


                • #9
                  With $GSO having 1139 ticks on the 2nd and 1550 on the 3rd, my gut is telling me that it is not because of the same pricing but rather because of ticks being dropped. Many years ago when Signal (a DBC company) was sent out via FM, I found a problem in the logic which was causing incorrect tick transmission, which was overloading the buffers and causing a loss of ticks. A couple of years later I solved some data problems with BMI and later with DTN. I have Beta'd market software’s for 15 years. I know my stuff.

                  Here is one example of at least 3 timestamps (that took at least 1 minute) on the 2nd for $GSO
                  5/2/2005 12:39:22 148.31
                  5/2/2005 12:38:07 148.29

                  It took 1:15 for the next tick


                  Assuming I am correct and ticks are being dropped, what action will be taken?

                  I ask because, before I spend additional hours to provide more evidence to what is taking place, I want to know the outcome of my efforts. I will verify the issue against other data vendors and get a capture from the exchange itself and "look" for multiple stamps with the same price if something practicle will be done.

                  Thanks

                  Comment


                  • #10
                    BIG1RED9,

                    We are currently receiving our data direct from the respective exchanges. In the event that the data that we are receiving differs from the data that is being broadcast, (i.e. missing packets and so forth), we are able to isolate any possible causes and make the necessary adjustments to correct this difference.

                    We are constantly running sanity checks for both the speed and integrity for the broadcast data. As Scott mentioned, we do not resend data when there has not been a change in since the previous tick. With this in mind, we are constantly checking and rechecking various markets at various times for the most complete sample.

                    If you do find that there is a discrepancy, we would take that to our engineers who would research this issue further. If a problem is found, it would go into a queue to be repaired.

                    Comment


                    • #11
                      Just what I wanted to hear....

                      I will order in the data samples and get to work and provide you with the facts as they show up. I have zero interest in being correct but a great deal of interest in the truth, no matter if I am right or wrong.

                      Thank you
                      Don

                      Comment


                      • #12
                        I received my data samples and went through about half a day's worth of tick data on $GSO. Once I detected a pattern I stopped.

                        As stated, the reason there is a difference between exchange and eSignal data is the exchange prints every tick, every 15 seconds no matter the price and eSignal does not transmit a tick, if that ticks value is the same as the one preceding it.

                        In the day I sampled eSignal had 1273 Ticks and my comparison data had 1565, so there are quite a few prints at the same value.

                        Don

                        Comment


                        • #13
                          BigRed,

                          Really appreciate the follow-up and confirmation of your findings.

                          Thanks.

                          Comment

                          Working...
                          X