Algorithmic trending strategies

Discussion in 'Automated Trading' started by abattia, Mar 24, 2011.

  1. “At best, algorithmic trending strategies tend to be breakeven.”
  2. thank you. the more people will think so, the better!
  3. If by "at best" you mean "on average" then this is true. There are losers, winners and break-even-ers. I mean the distribution has zero mean. Why is this something you have to ask a question about?
  4. Disagree .

    I could say that algorithmic RTM strategies tend to breakeven at best since I haven't seen one with my two eyes or come up with a profitable one yet. Just because I haven't seen one doesn't mean they don't exist or work right?
  5. Sorry. Apologies if the intended meaning wasn’t clear.

    No, the sense in which “at best” is intended is not interchangeable with “on average”.

    (In anticipation of much enraged flaming) I'll re-state the position more broadly. How about the following?

    “Algorithmic/systematic trading lends itself more easily to mean reversion trading. Algorithmic directional trading on the other hand is much more difficult and – for the most part – a given algorithmic/systematic directional strategy will be breakeven over the long term. In general, good discretionary traders beat algos/systematic trading strategies at directional trading.”

  6. rosy2


    you just post a topic and have no input yourself?
  7. OK, no problem.

    The quote is a rephrasing of a view expressed in an excellent trading blog by someone I have immense professional respect for (who I won't identify, to avoid this thread becoming a personal attack on anyone other than me!).

    I think the view is challenging.

    However, I don't personally have broad enough experience to assess its validity (although my own successful algos are indeed all mean reverters!).

    I hoped it might generate an interesting discussion here on ET.
  8. You can design an algo that has a 99% probability of success trading against a random number generator. The algo makes scalps of variable sizes 99% of the time but takes a huge hit the 1% it exceeds its limits.

    Statistically You have better than 50% chance at winning. The algo averages down so unless your trading is backed by the full faith and force of the US treasury at some point your going to hit limits.

    Now, if you are a broker offering to trade this strategy for clients charging a management fee, then you will always win.
  9. Hi abattia,

    sorry i am a little late at the party! :)
    (Today i had to rush some code changes a Canadian friend pointed out.)

    Here is the "guilty" thread: ;-))

    Well as many absolute statements out of context, anything can be true or false. I believe that, after a while, one realizes that making good strategies requires to put together different ideas.
    I do use a "component" within my approach which could be defined (by simplicity) as "trending", but i merge it within a countertrending context, as a way of hedging.

    ehi there are no personal attacks here. We are all here to learn and grow, and the mkt is a great and humbling teacher!
    I am always ready to backup and ride a new idea if it means a larger ($ / risk). That's what i actually do all the time! ;-))

  10. gsgs


    I think it did work in general before the advent of the computers.
    I once tested it on that daily Dow since 1900.

    Then many traders have been using it and it no longer works
    e.g. for the stock-indices or currencies.

    It may still work for commodities or smaller stocks, I'm not sure
    #10     Mar 27, 2011