Toxic order flow against limit orders

Discussion in 'Order Execution' started by digitalnomad, Jan 9, 2019.

  1. In reference to capturing price breakouts in liquid instruments, is there a set time in seconds anyone would suggest before canceling unfilled limit orders, to avoid toxic order flow? Non-HFT of course
  2. qlai


    You can't capture break out with passive/resting limit orders.
  3. Sure you can. I do it on a daily basis. To clarify, I’m not talking resting bracket orders. Liquid markets back and fill constantly. Once price meets the criteria, you send the limit order to add/take liquidity, hopefully leaning into the positive order flow. I currently allow 30 seconds till cancel, but toxic order flow is the one thing I can’t quantify, so just trying to gather opinions.
    Last edited: Jan 9, 2019
  4. qlai


    I'm sorry you are confusing me with terms. Toxic order flow refers to orders from informed traders. It's not measured in time. I don't know what positive order flow is. I guess it means the opposite of toxic. You are talking about adverse selection maybe? In any case you are in HFT territory. Hope you get better answers from others.
    GRULSTMRNN likes this.
  5. A grey area for sure. By positive, I did mean opposite of toxic. I think that’s accurate. It sounds good :)
  6. traider


    I don't think a general rule like cancelling after x seconds will work well. More likely you have to predict short term movements from order book data and this is in the realm of hft.
  7. qlai


    I can't remember which thread, but someone posted Dennis Dick's episode of Interview with Traders podcast. Very informative. One interesting thing he said is that he no longer posts limit orders and just takes liquidity instead. I can only imagine what will happen if they take liquidity rebates away ... Vaccum!
  8. sle


    With the prevalence of the HF players, providing liquidity exposes a non-HF trader to massive negative selection. It's obvious that in that environment (tighter spreads, higher potential for negative selection) it makes more sense to execute aggressively if you can't compete on latency.
    quant1, digitalnomad and qlai like this.
  9. I think you’ve convinced me to agree with you. This isn’t child’s play.
  10. This sounds interesting. I’m in:thumbsup:

    I’m presuming some might be building market making models around taking liquidity. You think so?
    Last edited: Jan 10, 2019
    #10     Jan 10, 2019