Fully automated futures trading

Discussion in 'Journals' started by globalarbtrader, Feb 11, 2015.

  1. FCT

    FCT

    Can you please start a new thread don’t pollute this high-quality thread with this junk.
     
    #1071     Oct 28, 2017
  2. https://github.com/robcarver17/pysystemtrade/blob/master/syscore/correlations.py

    GAT
     
    #1072     Oct 28, 2017
    AvantGarde likes this.
  3. My margin troubles have continued... basically I no longer seem to able to borrow against my largest ETF position, and IB generally are increasing margin requirements across the board. So I've decided to stop trading V2X and VIX as they are extraordinarily margin greedy.

    GAT
     
    #1073     Oct 28, 2017
  4. isotope1

    isotope1

    Is that just a CGT based decision? I would have thought it'd be better to dump the ETF and stay holding the VIX?
     
    #1074     Oct 28, 2017
  5. Yes, entirely down to CGT. Otherwise I'd have sold the ETF ages okay.

    GAT
     
    #1075     Oct 28, 2017
  6. Some time ago IB increased the margin requirements for US-based volatility products. This week they sent a message that they will now do the same for non-US volatility products. Until November 6th will the margin requirements for various such products be gradually increased. Besides the margin increase have they also found another trick to cover and protect themselves: for accounts which they consider "risky" will they start to charge a so-called Daily Exposure Fee. I have not yet been able to analyse this in detail, but I assume that positions in volatility products greatly contribute to this fee.
     
    #1076     Oct 28, 2017
  7. traider

    traider

    Hi GAT, was wondering if you ever looked at logistic regression, kalman filters or hidden markov models and found them useful. Logistic regression seem to be quite promising when trading 1 lot sizes.
     
    #1077     Oct 28, 2017
  8. isotope1

    isotope1

    I'm curious to know what you've been experimenting with. What have you tried? I've experimented with random forests, but got mostly overfitting.

    Just some background:
    With the trend following, the fitting of forecast weights is a linear regression. The kalman filter reduces to the EWMA in the limit that of random walk + noise, which is a reasonable assumption for this strategy. HMM I can't see how it could be that useful on daily data, although I'd like to experiment more.

    In other news, I'm trying to write market making software now, and all of these are way more relevant. If anybody here knows anything about market making, please let me know, as I know nothing! :)
     
    #1078     Oct 28, 2017
  9. SP/ES almost exclusively since it was trading 250 wow now 2500, others i have would include US30, Bund and anything that is super liquid and has aftermarkets, the more hours the better.

    I also dabble in bitcoin since it started and love that it is active 24/7. Maybe one day these Crypto's will become more liquid.
     
    #1079     Oct 28, 2017
  10. no problems spot light back on you.
     
    #1080     Oct 28, 2017