Good broker for automated futures and options trading?

Discussion in 'Automated Trading' started by snackly, Apr 6, 2009.

  1. How did you manage to come up with that nonsense about ES? Please read the available information from the CME about Globex technology before posting complete nonsense.

    The latest information (March 2009) from the CME about Globex speeds states that "We execute customer trades in under 10 milliseconds on average." That is a maximum of 100 price ticks per second.
     
    #41     May 2, 2009
  2. jjw

    jjw ET Sponsor

    the execuution time refers to latency not throughput. looking at the data i found that on April 13, from 09:30:00.000000 through 09:30:00.999999 the cme sent 182 trade reports for esm9.
     
    #42     May 2, 2009
  3. I believe that you are correct, but there is a direct relationship between trade report latency and throughput as trades cannot occur on a continuing basis faster than orders can be processed.

    The Globex technical literature published by the CME is less than precise in terminology and not well coordinated.

    The information I posted was from a March 2009 CME newsletter ("We execute customer trades in under 10 milliseconds on average.") and appears to be superceded by an older newsletter. On September 30, 2008 the following was published by the CME: "In Q4 2008, CME Group is implementing an upgrade to futures markets on the CME Globex electronic trading platform that will result in significantly reduced message response times. Already among the fastest in the industry, this upgrade is expected to reduce response times by an average of 40-50%, from an average of 10-12 milliseconds currently to approximately 5-7 milliseconds."

    At 5 millisecond response times Globex is capable of reporting 200 price ticks per second. Your 182 reported ticks in a second is within 10% of 200 ticks.
     
    #43     May 2, 2009
  4. jjw

    jjw ET Sponsor

    terminology seems to be imprecise. clearly the notion of the time it takes to execute a trade is vauge. suppose i send in an order that is somewhat away from the best bid/ask. the market takes a couple of minutes to get to my order and then it gets executed. does this mean the execution time was on the order of 2 minutes? i doubt it. i suspect that the CME is referring to the time it takes to process a market order or the time it takes to confirm receipt of an order.

    again looking at the data it is difficult to tell how many trades on april 12 at 9:30:00 to 9:30:01 were the result of market orders but we know it cannot be more than 182. on the other hand i noticed that there were nearly as many best bid/ask reports during that time so it could be that the cme can run at an average latency of around 2 - 3 milliseconds. if this is true then the differnece between latency and throughput is not clear.

    i can tell you from experience, especially with large market data rates (>100,000 messages perseond), there is a big difference between latency and throughput but at the edges the difference is quite blurred.
     
    #44     May 3, 2009
  5. taotree

    taotree

    Latency and throughput are completely different things. Something could have a latency of 100 ms and spit out any number of messages per second.

    If you go here:
    http://www.cmegroup.com/market-data/distributor/market-data-platform.html

    and choose Calendar, go back to Q3 2008 and you'll see MPS (messages per second) on the order of what I mentioned (10's of thousands). Now, that is referring to a group of symbols (equity futures) and not just ES, so perhaps I was overestimating a little (I was basing my statement on the "thousands of ticks per second" which I had heard (and appears to be accruate)--but possibly mistakenly applied to ES only instead of the group of CME equity futures). Also, I was not talking about trades/ticks per second only, but full market depth which changes far more often than just trades coming through.

    Also, Q3 2008 was highly volatile, so it could be suggested that this is highly unusual, and that's fine. But I was referring to peak and so... that was peak.

    http://www.a-teamgroup.com/article/...cs-for-january-peaks-for-nyse-nasdaq-options/
    Refers to new peak of 70k MPS over a year ago.

    http://news.medill.northwestern.edu/chicago/news.aspx?id=68273
    Refers to peak of 2k to 3k trades per second (NOT MPS) 2 years ago.

    It's likely the market has exceeded those numbers since then.
     
    #45     May 3, 2009
  6. jjw

    jjw ET Sponsor

    though i have not looked at the data for april 13, i do know that we routinely get more than 30,000 messages per second from the CME during the new york market open. and yes, order book updates per second far exceed the sum of the number of trade updates per second plus the number of best bid/ask updates per second.

    as a comparison, data from BATS can exceed 100,000 messages per second put the king of it all is OPRA: more than 900,000 messages per second peak, more than 600,000 messages per second sustained over a 1 minute period.
     
    #46     May 3, 2009