Ninja + Zenfire vs Ninja + TT feed

Discussion in 'Order Execution' started by hanzahar, Aug 15, 2008.

  1. RedDuke

    RedDuke

    I have also wrote it in another thread.

    Congrats to Zen-Fire, Mirus and AMP.

    I have checked the feed this morning, and its awesome. I have a demo zen-fire feed that I put against X_Trader TT feed (that is what I use to place trades) and the price and dom matched.

    But Zen-fire passes way more transactions in Time and Sales that TT.

    DOM refresh rates were identical for the most part.

    Finally zen-fir is a full working tool for Eurex traders.

    All my observations were for DAX futrues.

    I have a question: Is there any difference between zen-fire demo feed and live trading feed?

    Thanks,
    redduke
     
    #21     Sep 11, 2008
  2. Anyone knows some detail?!
    :eek:
     
    #22     Sep 16, 2008
  3. You are comparing apples to oranges.

    TT is not generic but tied to your broker and their IT infrastructure.

    ie. Velocity will have different price servers than advantage or another broker. Brokers that colocate at equinox should have the fastest updates from CME.

    TT is licensed technology that brokers can configured in various ways with many options on how to process and filter price feeds.

    Going back to the data center... Network distributions is a whole other game. Equinox for its managed services standardized on Dell and Cisco equipment.

    If your broker manages their own gear at equinox they can use faster bleeding edge technologies.

    Having a great feed does you no good if you can't execute and get filled. TT shines in this area as the fill servers and price servers are typically in the same rack. (~20 ms)

    ZF / Ninja is distributed technology which introduces latency for fills. (~100 ms depending on your brokers back office)

    Ultimately you need to evaluate the whole transaction process than just the feed.
     
    #23     Sep 16, 2008
  4. Big AAPL

    Big AAPL

    I trade the YM and although I expected a difference, there actually was none. Sometimes, I get a faster fill live than demo. Go figure - and yes, I am very happy with NT/ZF.
     
    #24     Sep 16, 2008
  5. moarla

    moarla

    >>PocketChange
    appreciate your more detailed coments on TT and ZEN...
    feed is different from execution speed...
    If you have any expirience, could you explain me the Difference in execution speed between TT ZEN and IB?
    I would really appreciate coments about that (but please, dont talk about feed, i know the difference there :)) )
     
    #25     Sep 16, 2008
  6. RedDuke

    RedDuke

    I have spoken to Mirus yesterday.

    They actually picked the phone during regular market hours, so I guess the things are running smoothly even in such conditions as yestarday.

    The guy confirmed that there is no difference, since there is no demo feed and we are seeing live feed in demo account, and the only difference is that there is no execution in demo obviously .

    Also, he said that once our transaction hits their servers, the execution takes about 20-30 mili seconds.

    Regards,
    redduke
     
    #26     Sep 16, 2008

  7. Can share some autotrading experience.

    Algo trading system using excel and processing level II price feed to trigger orders.

    Orders are triggered intra tick based on level II synced with our position sizes. ie.. bid/ask data is dynamically built for our current position sizes. ie bids and asks are different for 20 lot fills versus 1. Each change in quantity triggers recalcs and releases orders when depth supports the order size.

    #1. TT is bit kludgy.. the feed is rtd' into excel.
    Calculations are performed
    Cells are DDE linked back to TT autotrader client.
    TT Client places and manages orders with Broker TT order server.

    #2 Had to throttle order releases to 100ms to avoid double orders. Allow for time for fill data to update positions in worksheet.

    There is a lot overhead in retrieving and processing the TT order book. Round trip buffered to 100 ms for safety.


    Also using OEC... Different order book update mechanism.
    All orders are submitted and proceessed OEC server side.

    OEC DDE feed into excel,
    Orders are triggered and a provisional status is returned to excel by OEC client. No throttling needed but order confirmations take longer. 1 - 5 seconds. OEC is decent but certain aspects are outside of our visibility and control.

    Both have overhead but have sufficient tools to accomodate most trading systems.

    Ninja doesn't work for our trading system. No support for trading future options. We use future options to hedge positions.
     
    #27     Sep 16, 2008
  8. ok, great - thanks, B.
     
    #28     Sep 16, 2008
  9. andiamo

    andiamo

    For those of you that are using Ninja with either TT or Zen-Fire (or have in the past), what kind of network connection do you have? DSL? Cable? T1? Other?

    Also, for those that say that the TT feed is faster than Zen-Fire (or vice versa), how are you measuring this?
     
    #29     Sep 16, 2008
  10. RedDuke

    RedDuke

    I use cable, 35mbps. DSL is a backup.
     
    #30     Sep 16, 2008