Vectorized vs Event Driven Backtesting

Discussion in 'App Development' started by kojinakata, Apr 3, 2015.

  1. 2rosy

    2rosy

    #11     Sep 19, 2015
    tradingcomputer and nth like this.
  2. how does a vectorized backtest even make in the slightest sense?

    Example: Quote1 processed by Strategy1 triggers Order1 that is submitted and filled. The order manager processes the incoming fill and determines that currently no now positions can be established due to certain risk limits. Quote2 comes in, Strategy2 again triggers an order, Order2, which is however rejected.

    Can you explain how the above would be accomplished with a vectorized backtest?

    P.S.: You questioned in another thread that I work in this industry. I am curious how you run vectorized simulations/backtests even with the simplest of setups. Maybe I can learn from you.


     
    #12     Sep 19, 2015
  3. 2rosy

    2rosy

    What i wrote is a quick and dirty simulator. Its not vectorized. You or anyone can add to it to track a strategy's performance.
     
    #13     Sep 19, 2015
  4. Sorry I misread. Apologies

     
    #14     Sep 19, 2015
  5. Butterfly

    Butterfly

    what kind of crap was that ? are you bored or something to have so much time writing such pointless code ?

    pure amateurish work !!!
     
    #15     Oct 25, 2015
  6. Apologise for my ignorance...

    Just would like to confirm my understanding.

    Vectorized backtesting can deal with a single sequence of related events, so it works for a single path strategy only. For example, trigger order 1, wait for order1 status, after 1 minutes if order 1 not is filled, cancel order 1, then submit order 2.

    Event-driven backtesting can deal with multiple inte-related sequence of events that are running in parallel and may have dependency to other events. For example many inter-related orders are executed simultaneously.

    Is this correct?




     
    Last edited: Nov 19, 2015
    #16     Nov 19, 2015