What type of tick data is best for historical simulation.

Discussion in 'App Development' started by 931, Aug 1, 2020.

  1. guru

    guru


    Yeah. I mean “fake” in the sense they can be for small qty, or have hidden orders and larger qty in front of it. Lots of scenarios.
    Though I don’t trade at high frequency, even limiting some strategies to wait X bars before closing a position, to make sure I am predicting something vs trying to game small price fluctuations.
     
    #11     Aug 1, 2020
  2. My experience is that this would be too optimistic. Often your fills will be worse than the quote data. With "worse" I mean that your losses will be larger and your profits smaller.
     
    #12     Aug 1, 2020
  3. 931

    931

    I use have used nbbo quotes for backtester so far and can add extra slippage on executions.
    Probably could get slippages more accurate with more live trading logs...

    Atm i dont consider quoute bid and ask volume data at all , only bid & ask prices.
    And i allocate <5k per stock on live market.
    In some cases maybe quote volume data would make simulation more accurate and create slippage or half fills but my account is too small for that in most cases probably.

    Could nbbo quote volume data be still beneficial in some way?

    Also i find it hard to understand quote data condition parameters. Not much info on google. So i am not sure on my data processing accuracy.

    With trade data there are gaps sometimes and it might slip alot on sl or tp... But i suppose it may be more realistic to expect slippage as mentioned.

    Id like to also work with penny stocks where trade data could be gapped alot.

    Ive read that some people make algos to sync & utilize both trade and quote data in backtests.
    But i dont fully understand the benefit it could give in my case.
     
    Last edited: Aug 2, 2020
    #13     Aug 2, 2020
  4. 931

    931

    Thanks for pointing it out , i suspected same .

    In quote data documentation near volume descriptions it says obligated volume.

    But it can be interpreted many ways.

    Could it be so that the conditions change on last microsecond? Or some predatory HFT has info about coming order before order gets executed and pulls it away before...
    Rendering quote data inaccurate...
     
    #14     Aug 2, 2020