Low-latency data feed and execution APIs

Discussion in 'Automated Trading' started by loopquantumgirl, Jul 20, 2013.

  1. Which low-latency data feeds and execution APIs would you guys recommend for a 6 to 7 digit annual budget for infrastructure? I'm just being short of the capital to write feed handlers for every exchange directly although I'm open to compelling arguments to do so.

    1. I hope to cover these markets in decreasing priority:
    - Execution and data in all U.S., Eurex, SGX, KRX, HKEx, ASX (SFE) futures.
    - Execution and data in Tokyo, HKEx, KRX, SGX, MYX equities.
    - Data in all U.S. and LSE equities.
    - Data in above-listed markets' futures and equity options.
    - Data in OPRA feed.

    2. I'm looking for DMA; if it's a broker-specific solution, the broker should not be receiving any payments for order flow.

    3. Lowest latency is not my priority. Max 120 microsecond RTT in U.S. markets is the service level requirement. (Adjust appropriately for the other markets.)

    I've heard TT via Newedge, ORC and QuantHouse so far. What about data?
  2. Sounds like Bloomberg EMSX (comes free from with the subscription), for $2000 a month with a 2-year contract (or $1800 for multiple subscriptions) , could be a decent solution for you. Co-location for this though would be dependent upon the broker-end solutions or what you would be able to setup manually or with Bloomberg directory (http://www.bloomberg.com/enterprise/enterprise_products/hosting_solutions/).

    I believe that Realtick has also recently co-located with NY4 and they also have a data-center solution in Chicago, but that's for US markets only.

    Both platforms are broker-neutral, multi-asset and offer market data + execution capabilities. Both have FIX APIs. Realtick uses Interactive Data for international markets data, but uses its own data for the US.

    What are you using to store Tick Data BTW?

    TT and CQG are futures only, although some brokers will let you trade hotspot through TT. QuantHouse isn't for execution (expect not even microsecond, but millisecond latency), they purchased that QF product from a retail vendor, but they have since relatively stalled development on it. It's mainly a backtesting solution. They're focusing on data now.

    ORC or RTS or Flextrade can do what you're asking on the execution-side (RTS and Flextrade also have very decent risk management solutions), but then you need a market data solution.
  3. Also received another recommendation for RTS (http://www.rtsgroup.net/). Would be nice to have some color on them.

    Cool, thanks! I don't like Bloomberg for any means, but will look into Realtick. We employ a tier 1 solution from EMC. The software implementation itself is pretty standard.
  4. hoppla


    120 microseconds round-trip latency on CME and/ or Eurex Futures? I dont know ORC or RTS but highly doubt this is achievable especially going through a 3rd party technology stack. I'd say best you can do is maybe 1.5-1.8millisecs average RTT (ie sent to ack) on CME for instances using TT, RTS,...

    Unless you mean 120 microseconds RTT from/to the 3rd party's execution and pre-trade risk servers from/to your box?
  5. I think ORC is similar to slightly slower than Tbricks (founded by former ORC guys, around 800 microseconds), but Flextrade has around 20 microsecond latency and RTS can do 100-150, if you use their co-location services.
  6. hoppla


    Again which latency and exchange are we talking here? I am specifically referring to CME and the roundtrip time it takes from a colocated box in Aurora or Equinix/ Cermak (will be a bit higher) to CME's matching engine and back. I haven't measured this recently, but last time I checked it must have been around 2ms using an all 3rd party non-optimized technology stack. This seems to be in the ball park when comparing with CME's own numbers (page 5) here:


    It's from May 2012 so a bit dated and I am sure they have improved since but doubt down to 20 microseconds.
  7. Those numbers represent an aggregate that includes ordinary orders, it's not a measurement of ultra-low-latency connectivity.

    The B2bits FIX engine, for example, is pre-certified with the CME, these are their numbers:



    "Trades in those contracts now occur, ON AVERAGE, in a range of 10 milliseconds to sub-3 milliseconds depending on volatility or trading matching dynamics"

    Also, see:


  8. hoppla


    That is not a loophole (referring to the article's headline). Of course, you will receive fill messages when your limit orders are hit before they appear on the public feed. The link you provided:


    as per my understanding just refers to the DMA API part of the technology stack and not the whole round trip latency. In particular the figures stated in regards to their iLink Routing abstraction of the underlying fix protocol refer to internal latency experienced on your box:

    Meaning their fix engine takes up to 16us to interpret a message and to fire the corresponding callback into the subscribed client and vice versa. Have you measured your total roundtrip latency as in the time it takes for your order placement message (talking limit orders here to be specific) to leave the box's network card until you get an acknowledgment from the exchange that your order has been received and been added to the book? I've never looked at market orders' latency as it's not really relevant to the systems so I don't know about that one.
  9. #10     Jul 21, 2013