Where does the automated retail trader start to find edge?

Discussion in 'Automated Trading' started by garchbrooks, Feb 22, 2010.

  1. Oh, with one exception: trading the news with the herd. But this is still most of the game of the manual trader, more so than the automated trader -- the vast majority of retail tools don't allow widespread portfolio-level testing in such a way that you could backtest a strategy that trades news-based signals across lots of different stocks (or maybe even futures.) So I don't think this class of strategy is as effectively used by automated retail traders, although feel free to correct me if I'm wrong.
     
    #21     Feb 22, 2010
  2. You could do this.

    I use Sierra, it would be complicated but it is possible.

    There are most likely much easier ways of doing this though and it would most likely take a team of people. If you have not designed a system before it really is quite an undertaking and I recommend if you haven't - just don't.

    Save yourself the 100,000 hours of hell.
     
    #22     Feb 23, 2010
  3. Yes and no. My experience tells me there are no-news spreads, which are tight and efficient and then there's vapor-spreads, where you "see" (i.e., at a human speed) a spread, but your fill at that spread is unattainable. Like the other poster said, if it's an illiquid stock, that vapor-spread will wack you hard and remove the edge (like the 6 cent one in the article).

    If you talk about a trade that's obvious, it's always always a vapor spread to the retail guy. If the trade is not crowded, you go back to the problem of retail trader risk being extremely high. So this goes back to the original question of how these ET retail guys are consistently profitable, if at all.

    Crude technical analysis trades are not workable in the long run without a lot of subjective tweaking. Yet, from posts on ET, I see that tradestation and IB users on the whole are actually profitable.
     
    #23     Feb 23, 2010
  4. Still man I don't know, a lot of assumptions. You know what that means.

    Most technical indicators are garbage IMO, but you need to find out what works best with your methods. Trial and error and a large enough sample size will be what takes your programs to the level they need to be on.

    I have designed 0 programs for equities anyway, all based on Futures and FOREX, but mostly futures. Not a real fan of equities personally. Maybe you should try and change your game up.

    GL
     
    #24     Feb 23, 2010
  5. jjw

    jjw ET Sponsor

    Our tests were conducted while getting approximately 2000 market data messages per second. Our release criteria was simple - upon receipt of every nth market data message, release another order. A trading algorithm would likely take a bit more time to decide to release an order but probably not too much more time.

    We have removed the latency of the user's network by putting all programs on machines in the data center.

    In addition to letting you know about our new software, my point was that the entry barrier to the HFT game is now much lower. No longer does one need to commit millions of dollars to do this and no longer does one need to have an extensive staff to handle exchanges, networks, etc. For relatively little dough, a serious trader can enjoy the benefits of the same technology that had been available to only a few.
     
    #25     Feb 23, 2010