blackbox algorithms for interest rate futures

Discussion in 'Automated Trading' started by TCM, Jul 2, 2006.

  1. TCM


    can anyone provide some color on the types of automated/algorithmic strategies that are currently being used in eurodollar/euribor/treasury/bund futures? are they all essentially some form of scalping that utilizes order book dynamics and market microstructure?
  2. why would you like to trade in particular for eurodollar/euribor/treasury/bund futures ? I am not an expert but as I know we can simply brute force the strategies with respect to the intraday data for any edge, no matter what kind of investment it is. I encounted a problem on the process, called backtesting in the context of automated trading, requires extensive computational power, say, simulating a single strategy on treasury with 1 month of intraday data takes 8 hours. So, it is good that you ask for the correct strategy before backtesting it. This is the case for scripting based backtesting platform. However, Lawrence Chan, the developer from Neoticker told me that the strategy part can be written by compilable language for much faster execution and thus the backtesting might takes minutes then. You can then bruteforce the strategies. Formers, please correct me if I am wrong.
  3. TCM


    i'm specifically interested in the fundamental "strategies" at a high level that are being written, and not so much the details of the code or backtesting. what is the "logic" behind the algorithms for automated trading in these STIR and bond futures on cme/cbot/eurex/liffe. you see very obvious activity in the order book such as 25,000 contracts on the bid to take advantage of the pro rata filling in eurodollars, and was just curious how people are using automated scalping via machine algorithms?
  4. Only losers will provide you with ample ïnformation
    Winners know better.
  5. Perhaps try searching for "Spread Trading".
  6. Pabst


    There's always one side of a spread that's stronger or "better". Knowing what you do about the pro rata allocation you can figure the rest........
  7. I know someone who trade 30-year bond future ...

    mechanically over the past 10, well it adds up to 15 years now.

    The system scalps 3 to 5 ticks every time.

    As far as I know, the model worked all these years.

    No order books, no special data - purely tick data analysis (with bid/ask data).

  8. wow that is slow.............have you ever looked into ?

  9. Yes there are very good strategies and you do not need a single algorithm to have success.
  10. Isnt it all proven strategies base on tick data analysis ? Could you tell me anything we can know on the order book or others for competitive advantage ?
    #10     Jul 2, 2006