Interactive Brokers data quality

Discussion in 'Data Sets and Feeds' started by MarketMasher, Dec 22, 2011.

  1. I've received a few pics lately from someone who asked me to cross-check his data (from IB) vs mine (unfiltered tick) and I have seen that his IB charts feeding NinjaTrader are sometimes missing the high or low tick of a swing (minute charts). When I asked him to refresh his charts, the missing low or high tick of the bar swing bar shows up.

    Does anyone else have this issue with IB live data? He isn't running too many futures charts, so I don't think it is a load problem.
  2. Everyone has it.

    Live data packets are sent a maximum of once per 100ms (used to be the rate, might have changed) and contain the last price in that period. If there is a price within that 100ms that exceeds your bar/swing high then you'll miss it.

    The advantage is that you don't get lagging data in fast markets and there is less congestion.

    IB created 5 second data to overcome the loss of information in live data. It sends a ohlcvt bar at the end of each 5 second period.

    If you look at IB's charting they combine 5s and live data to give a responsive chart that is corrected for lost data. As of version 920 though they still had some issues as I noted some bars shortened (they should always get longer as the period progresses) early in the bar's development.

    I created a function for Sierra Chart that combines the two correctly (without the shortening bar problem).
  3. Thanks for that. Nice to know that about IB data.

    I've known about the bundling that makes IB data pretty much useless for many tick and volume indicators. Now it seems like even their minute data has problems too unless addressed through programming. :p
  4. Actually - can you clarify that answer?

    Because if you are going to miss a high or a low tick in 100ms, how are you going to not miss it in a 5sec interval?
  5. Specterx


    Bottom line, IB is a brokerage not a data company. I would never use them as my primary data source.
  6. Because the 5 second part is geenraetd not from the same data. It is generated from IB using the ticks, then sent every 5 seconds (if I read that correctly). As such it contains possibly data tha the "ticks" do not show.
  7. +1. i suggest you use ib for execution and either kinetick or iqfeed (kinetick is just iqfeed rebranded but it's cheaper if you're using nt) for data. don't be put off by spending the extra $50 or maybe $100 a month on data b/c by definition is represents what you're trading. if you're looking at the wrong data how could you possibly trade right? if you don't trade off tick/vol charts like another poster said it doesn't matter.
    Erv Jus Soli likes this.
  8. Thanks for that.

    I have already suggested that he use an unfiltered feed, but he is starting out and wanted to watch the market for a while at a low cost.

    I never knew that even the minute data was unreliable from IB, as I figured that is the lowest denominator for data so sending 1 minute bar data at the close of the bar would mean the OHLCV of.....1 minute......

    Guess not (for IB).
  9. The charts you see in TWS which are created by live ticks coming in are not accurate. This means if you open a chart and wait for 5 hours, the 1minute bars will not be accurate. However if you load the same chart from the IB historical data servers by reloading the chart, it will be much better, but still not perfect. This is because the historical charts at IB are constructed with less missing ticks as they use their own full data feed to construct those. However, even the historical charts will have a few highs and lows (and even closes) missing a bit, compared to other data vendors.

    I've compared them to esignal and for the SPY the historical 1minute charts were missing around 1 in every 20 or so bars by one tick on either one of the OHLC values of the bar. So if you take eSignal historical data as benchmark, that's not so great.
  10. BSAM


    Oh, my.
    So one can't trade profitably using IB data?
    Wow...Who believes this?
    #10     Dec 23, 2011