random entry trading

Discussion in 'Trading' started by rosy2, Jan 23, 2007.

  1. rosy2


    many, if not all, traders make trades based on some analysis prior to entry. has anyone come across studies of trading based on purely random entry and the strategy is exits only?
  2. Simply, strategical entry / random exit equally works.

    The thought of random entry / strategical exit... is pretty much an urban legend towards "Exits are more important"

    They are equally important. Over-emphasis on one side of the trade is immature...


    random entries have been discussed many times over many years. Random with money management is'nt an edge.
  4. very interesting idea. ignoring the problem of spread and commission for a few minutes. a random entry signal should have a 50/50 chance of being a profitable trade. with an expected profit tending to 0 over time. now take a simple exit strategy. say a -1 bip stop loss and a 10 bip take profit.

    i think over back testing you would find that running this strategy will not result in a 0 profit expectancy. based on my experience, a random strategy will tend to have a negative expectancy over time :p still working in the no spread / no commission dreamland. if we do find a negative expectancy logically we should be able to flip the exit strategy, 1 bip take profit and -10 bip stop loss, the expectancy should also flip, assuming the market is a random distribution over a long enough period of time.

    so the challenge here is to make the negative or positive expectancy (of an exit strategy running with no spread / commis), greater than the cost associated with the spread + commission.

    someone want to enlighten me where the flaw in my logic is?

    (yeah i know you cant prove the market is a purely random variable over time. but even if it is not. a purely random entry strategy should balance the number of trades with the market bias against the number of trades not with the bias.
  5. I was thinking more about this idea. Is the problem that because as you grow your # of trades the chances increase of a streak that will close out the account?
  6. You can't consistently win a coin flipping game over the long run, no matter which strategy or money management is employed. It's that simple. And if you add transaction costs (such as spread, slippage, and commissions) to every flip, you are guaranteed to consistently lose.
  7. telling me my logic is flawed (i already know this) is not the same as telling me where my logic is flawed. at which point of my example do i make an assumption that is false is really the question i am asking.
  8. So, you already know where the flaw is, but want to see if the others can spot it? Is this a homework assignment of some sort?
  9. i have no idea where the flaw is. from a mathematical stand point, it 'seems' sound . given a 50/50 distribution. no cost. equal pay out in either case, say stop loss at -5 and take profit at 5. the expected profit should be 0 over a long enough time. i know based on experience that it is fairly easy to screw the expected profit in the negative direction by playing with exit conditions. based on the math you should be able to invert the expected pl from profit to loss or vise versa by inverting the exit conditions.

    i assume what you are saying by stating 'it can not and will not ever work' is that it is impossible to get a negative or positive expectancy in the 50/50 no cost case that is greater than the costs associated with trading? I find this statement hard to believe without some kind of reasoning presented with it.

    im not taking a poll on if people believe this is possible or not. im looking for someone who has done some research in the area or has spotted the problem an assumption in my math to respond.
  10. Is the entry purely random? are the exits purely random?

    these are the critical question to ask.
    #10     Jan 24, 2007