NQ 1min v. Tick data discrepancies

Discussion in 'Data Sets and Feeds' started by joe flapjack, Apr 6, 2002.

  1. - Data provider is E-Signal
    - Charting is through Metastock 7.2

    I am playing with yesterday's (Friday 5 April) NQ charts and found numerous data discrepancies between data shown on my 1-minute chart and data shown on my tick chart.

    An easy example is the 8:40am 1-minute bar which shows a high of 1419.50 while my tick chart shows nothing above 1419.00 for the whole day. There are a bunch of these discrepancies.

    I have pulled local data as well as loaded fresh.

    Is this an E-Signal problem (and if so, why by changing time frames does the data changes)? MS problem (and if so, why or how does MS 'interpret' the same data incorrectly)? or ?



  2. stevet


    i keep reading on here that everyone thinks esignals data is great - but i found a lot of anomalies in their data, which ultimately they end up agreeing is the case, but they never get back to you with why

    half the problem is that most who take esignal dont really understand data delivery, and the implications of candles etc, so they dont see the problems and therefore dont complain, so there is no pressure on esignal to act

    if their data works for you and you are making money, dont look too closely at their data - it might affect your sleep!
  3. Good points -- both.

    Yup, wanna know "why" and the data is making me money, so I'll think I will take a deserved, restful nap :)
  4. tntneo

    tntneo Moderator

    stevet, you are probably right.
    However, if you need to be sure about the candles there is much more trouble waiting for you.
    2 different data providers and charting software will not show you the same candle on 1 mn bars. that's a fact.
    It's going to be hard to find a data provider that satifies you then. maybe CQG.. what do you use ?
    And then again, your internet connection will change the candles too because of timing, lag, latency.

    I am not defending esignal. I said right away, you are probably right about them. I only one to point out that on 1mn bars many things can affect the candles, not just the data provider.

  5. stevet



    that was my point - the implications of data delivery and candles is something you understand, but most dont - and i dont use candles for all the doji stuff, just as an easy way to see h,l,c and therefore the implications of relative short term closes around key support resistance levels in futures and cash

    its one of the reasons for using cqg, if only because a lot of proffesionals use it and therefore you are seeing and acting on what they are seeing as well - even if it is wrong!
  6. Curious:

    I just downloaded the T&S table from E-Signal and it *does* show 1419.5 (unlike tick chart in MS).

    Looked at E-Signal 1-minute and again, 1419.5 is shown.

    As both TF in E-Signal show 1419.5 and only MS (in tick chart) does not, is this then a MS problem?
  7. Joe -

    I'm looking at the NQ data (on Quote.com though) and it shows a 1419.50 trade at 9:39:16am ET which is probably what you're seeing.

    If Metastock isn't showing it, it's probably a problem with it rather than the datafeed. Were you connected to the same physical server when you looked at it in MS vs. the normal eSignal program? if not, it could be a server-specific discrepency. If so, then it's probably an MS bug.

    Re: candles and datafeed - YOUR datafeed won't affect the data contents of the candles. Even if there's a slowdown in your datafeed, it's not going to change the contents or times of the candles shown on the chart.

    However, THEIR (the data providers) datafeed and internal clocks will affect it. If their primary datafeed over which they receive the exchange data is slowed or congested it can cause their servers to timestamp the trade data late and construct the bars incorrectly using the improperly stamped trade data.
  8. tntneo

    tntneo Moderator

    it depends, your internet connection (and your internal clock) can participate in building the wrong candle. because some charting software (I was not only talking about esignal, but esignal may have this problem with third party charting software using this datafeed and others) will use your internal clock as opposed to the time stamp at the server side (which, as archangel explained, might also be wrong).

    many, many charting package are not reliable for 1mn candles. that's the untold truth out there. few realizes that and most don't care (I sure don't care) but I am fully aware though so those who care should study the exact mechanism used by their providers.

    (esignal timestamps are correct ONLY when you reload the data from their server AND if the data was eventually corrected. you can't be sure of realtime data being broadcasted if you are using esignal data in your own charting package, unless you are sure of their implementation. the engineer in me is brought back to surface too quickly here.. scary.. let me slap myself)

  9. stevet


    joe flapjack

    there are also some issues in esignal data in metastock - one of which is that they use different timing - i can't remember which is which now - but one ends the interval on the stated time, and one starts the interval at the stated time - there is also the fact that metastock compares new data to any old data stored, in order to update any changes - and this can cause some anomalies, and certainly throw metastock into a loop

    if i recall correctly, they also build there interval periods from different time frames
  10. I use the eSignal data feed in Metastock Pro for realtime trading of the eminis. The realtime data and subsequent trading signals will not compare to the refreshed data due to corrections that occur and that are updated in Metastock. Of course the refreshed data is the corrected data, but that does not help the matter when trying to compare realtime signals with the after-the-fact signals with the corrected data. I have compared my realtime signals vs. the corrected data and the results seem to all wash out about the same over a given period of time.

    I end up not refreshing the data, unless it is an absolute must. If the refresh(on Metastock) is used, you will often see signals that were there disappear or signals now appearing that were not there before the refresh! :(

    I am using one minute data. This one minute data that is maintained in the Metastock realtime data cache is used to determine 5, 10, 15, 30, 60 minute bars for charting purposes. You can maintain about 65,000 one minute bars of data in the cache which is close to 7-8 months of regular hours trading data. You need to remove the afterhours data in order to maintain this much data however.
    #10     Apr 6, 2002