Automate Scalping

Discussion in 'Automated Trading' started by Norm, Nov 30, 2005.

  1. Norm



    Does anyone have any positive experience with automated scalping. If so, is bid and ask data an important aspect of the method?

  2. Bear in mind that according to popular ET wisdom, 95% of the answers will come from losers.
  3. I have spent a lot of time trying to develop high frequency systems that make consistant profits and to be honest I have never had any success. I have generally found the larger the time frame the more profitable the system.

    My profitable systems seek to identify the primary direction for the day and capure a large portion of that trend.

    I think that beyond 2 or 3 trades a day per market it's difficult to remain profitable with auto strategies. This is because you need to occasionally capture a really big move to make up for many little loses. If you're trading high frequency, you mazimum win size is too limited. For high frequency to work you need a high win rate which I have never been able to achieve.

    This is just my experience, I'd be glad to hear from anyone that has built a profitable high frequency system.

    A few of my high frequency systems used bid ask info but I don't believe it offers much for the following reason. You can have big volume smashing the bid, but if you have value players accumulating massive long positions, they'll just sit their and absorb all the selling pressure. It means they are getting a better price on their long term position.

    A high frequency system that uses the bid/ask data will have you short in this situation and you'll be on the wrong side of the market.

    So if you're going to utilise bid/ask, then I suggest you use it as a filter to time entry, once your underlying strategy has determined the primary trend as well as likely support and resistance.

    And generally once you identified the primary trend and likely S/R, the bid ask is no longer so important, because it tends to add noise not remove it.


  4. Interesting Bear, what do you consider high frequency? I think it is possible to achieve profitability with a high frequency system, but you are absolutely right, a high win rate is a must. For me a system that trades 100's of times a day and captures 2, 3 or 5 cents with a hard 5 cent stop can work. Obviously you need to have a good win ratio, especially when you factor in commissions, but it is doable. I am currently involved in developing such a system and would appreciate your feed back.

  5. I consider high frequency to be more than six trades a day.

  6. Interesting, I run a couple of automated trading systems that generates at least a few hundred trades daily (more than a thousand if all the systems together). But my algorithm is "pseudo market making", which requires a lot of trades (by definition). Actually my win % (if there is such a thing) is quite low, a lot of "push" trades, naturally commission and other fees (exchange, regulatory) accounts for quite large portion of my gross profit.

    And my system is not even the busiest trading I have ever seen. I have seen automated systems that accounts for as much as 8-10% of the daily volume for a given product, now that's very serious black-box trading.

    Naturally, none of the system will be that profitable in the retail range, the systems are all run as professional or institutional ATS.
  7. Interesting, I run a couple of automated trading systems that generates at least a few hundred trades daily (more than a thousand if all the systems together).

    Ya that is more along the lines of my idea of High Frequency Rufus. Sorry Bear, but 6 trades a day in my world is like being asleep. Anyways, as you say Rufus, 'push trades' or scratch or flats as I call them are quite common. In fact in our our system we use trailing stops that with often produce this result if a stock is highly volitile. Commissions are not a serious issue for me as we have the proper deal structure in place to facilitate a high number of trades and many breakeven trades.
    The system I am referring to is not exactly a market making type of box application but it does attempt to capture spreads and pick up liquidity credits whenever possible. This also helps to offset the trading costs. The true key obviously is to protect against large losses at all costs.
    Out of curiosity what would you say is the average win ratio of your system, 60% or better?
  8. He asked me for my definition of high frequency, not an agreed definition.

    It's fair to say that any strategy that attempts to make money without reference to an underlying directional trend is likely to be high frequency, simply because the more trades (volume) you push into the market, the more money you make. This to me IS market making - which is the highest trading frequency of all. For the retail trader though, this is not worth worrying about. You need cubic metres of money to make this stategy work.

  9. Much lower than 60%, 50% or more of the trades are push trades. Obviously commission and fees come to play here.
  10. Havin spent some time with market making firms, you will be surprised how small capitalized some of the MMs are. I have seen small locals with < 1M, which is reachable for some retail customers.
    #10     Dec 2, 2005