Announcement

Collapse
No announcement yet.

Partner 728x90

Collapse

Bulk Level 2 removals

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

    Bulk Level 2 removals

    Hi,

    Using FGBL 12-10 at 21 Oct 2010 09:44:05 (GMT+1), downloaded replay data.

    There is a massive bulk removal of ask and bid - about 20 of each, all at position zero. The reinsert comes 4 seconds later. The DOM doesn't do any of these removes. Does the DOM have special code in to detect this sort of situation? (I can see why it might)

    Thanks
    Dave

    #2
    I've looked further into this. It happens at least a few times a day. Isn't this an error with Zen Fire L2 data? Clearly the reality should not be to remove 20 bids in a row, or have I misunderstood what L2 remove position zero means?

    thanks
    Dave

    Comment


      #3
      Dave, thanks for the report - what are you comparing the DOM output to exactly? I'll need to clarify if / when the DOM would do any filtering, but the I suspect it would just pass on the Level 2 data received.

      Comment


        #4
        Thanks for the reply. I'm printing out the level 2 from a strategy. That's where I see all the removes.

        Comment


          #5
          Hi Dave, unfortunately we could not disclose how our DOM is constructed internally nor could we provide analysis why specific events are in the datasets. We do record every change change in Level 2 data that is submitted from the provider to us and the data works well with our reference sample showing how to build a price ladder using the OnMarketDepth().

          Comment


            #6
            That's really a bit of a cop-out.

            OK, let me try a different approach. Your DOM does not match up with Level2 data. This is serious. In the example given there are at least 20 level2 events that aren't reflected in the DOM at all for FOUR seconds, and are never actually reflected in the DOM.

            regards
            Dave

            Comment


              #7
              Dave, this is not a cop out - but the considerations that apply working with you on that topic - we could not provide analysis of the dataset events or how our DOM would use them.

              Comment


                #8
                Of course it's a cop-out. Just tell me that it's expected behaviour and why and I'll leave it. I don't need to know anything about how your dom works.

                If you can't tell me it's expected behaviour then the dom isn't working. I'm reporting a bug with the dom.

                As professional traders start to try and use NT7 for proper automated scalping, they are going to be asking support endless questions about your handling of level 2 and bid/ask. They aren't going to be happy with 'we can't discuss it'.

                Comment


                  #9
                  Let me try yet another approach.

                  If you use the sample level 2 book that you suggest as being

                  "data works well with our reference sample"

                  then you get an empty bid book and an empty ask book, at the same time as the order tab shows a normal bid/ask. The order tab should show N/A. That is clearly inconsistent.

                  So, I deliberately didn't mention the dom.

                  regards
                  Dave

                  Comment


                    #10
                    The order tab runs from the L1 data stream and not L2 data stream, cannot compare the two. As for the DOM, the short answer is that the DOM is coded differently and does not run off of removal events as our L2 window would, this is for performance reasons. There was an issue with B23 and earlier where removals were not reflected in the DOM, this has been corrected with RC1.
                    RayNinjaTrader Customer Service

                    Comment


                      #11
                      Thanks for the info Ray. I suppose the answer then is to ignore level2 removals (although I'd love to know what the whole order book being removed means in reality) and take latest bid/ask and bid/ask volume from level 1, then all other level 2 from level 2 data.

                      I can see how that should work fine. Only problem is it's impossible to test except on live data because your level1+2 replay files don't mesh to less than a second.

                      regards
                      Dave

                      Comment

                      Latest Posts

                      Collapse

                      Topics Statistics Last Post
                      Started by Geovanny Suaza, 02-11-2026, 06:32 PM
                      0 responses
                      599 views
                      0 likes
                      Last Post Geovanny Suaza  
                      Started by Geovanny Suaza, 02-11-2026, 05:51 PM
                      0 responses
                      345 views
                      1 like
                      Last Post Geovanny Suaza  
                      Started by Mindset, 02-09-2026, 11:44 AM
                      0 responses
                      103 views
                      0 likes
                      Last Post Mindset
                      by Mindset
                       
                      Started by Geovanny Suaza, 02-02-2026, 12:30 PM
                      0 responses
                      558 views
                      1 like
                      Last Post Geovanny Suaza  
                      Started by RFrosty, 01-28-2026, 06:49 PM
                      0 responses
                      558 views
                      1 like
                      Last Post RFrosty
                      by RFrosty
                       
                      Working...
                      X