generic strategies

Discussion in 'Automated Trading' started by kotika, Oct 3, 2005.

  1. kotika


    I still find the discussion on this forum lacking very much in the discussion of strategies. I dont think it is so very precious to reveal the generic types of things people might do. Anyway, there are only two basic generic strategies that i know of, and i doubt there are any others. The real edge is in choosing the best instruments to trade, the quality of the historical databases to tune parameters, and all the rest of techno stuff that people on this thread seem to obsess too much, like latency, execution cost, computers, efficiency of GUI etc...

    With that, i will try to list the generic strategies below

    type 0 : real arbitrage, buy an identical thing in market A and sell in market B

    type 1 : mean reversion, if A and B are interrelated, you can find trades that take advantage of their causal relation. For example if A sinks, you might buy A and sell B short. You can win both ways, if A was wrong to sink it might recover, or if things are so bad B might follow and sink too. This is bread and butter to hedge funds, and everything i personally do in all the various markets falls into this category.

    type 2 : trend following, this where you have some technical indicators to tell you to get in just when the move is confirmed to have started. For example, buying when short term MA crosses longer term MA, when price breakouts, and a zillion variations on this same theme. lot of futures traders do this sort of thing traditionally, and i'm sure many are successful, but i am not a fan.

    How about we take a poll, and have people post if they are primarily doing 0, 1, or 2. If i you are not sure ask for clarification. If you strongly believe what you are doing does not fall into any of the above, i would like to hear that too, in my 11 years trading i never heard of something that doesnt fall into one of these cats.

  2. mahras2


    My trading model is basically 80% mean reversion 20% trend following. The trend following concept is key because even though markets tend to trend 10-20% of the time the magnitude is great. Thus during periods like that the model uses that input and creates a directional bias in that direction. However in the 80-90% of the time markets range the model makes it many trades trying to capitalize on the fluctuation. Seems to work well for me.
  3. 3. Mean reversion on a single instrument? Selling overbought signals and buying oversold signals.
  4. How about an integration of 1, 2, 3, 4, ... etc.? :confused:
  5. kotika


    indeed, you can do mean reversion on a single instrument, if you dont have the means and ways to hedge your bets. this is not a strategy but inferior version of what i called type 1. sure i used to do it myself, but the homework required to do hedged trades is well worth it.

    second comment, before you suggest combining 1,2,3 and 4.... you may tell us what 3 and 4 is.
    maybe you can write a single piece of software that does all things separately, but i dont see how can you combine trend following with mean reversion at all?
  6. mahras2


    Well I do it. Without compromising the specifics here are some stuff I have used to combine elements of both:

    Typically mean reversion works due to the fact that typically markets remain at a tight trading range for 80-85% of the time. However, trends although only occuring 15-20% of the time can be exponentially greater in magnitude. Areas when mean reversion fails is when it keeps on placing trades against a trend thus continuously losing as the trend progresses. What I do is use my trend systems to create a bias which is incorporated into the mean reversion system. Then the mean reversion method determines on which way to shift the trades (long or short) based upon the data fed by the trend observer. Seems to work for me.

  7. This doesnt make any sense.. mean reversion is the opposite of trend following.

    Sometimes one works better while the other doesnt... its the way markets work...

    To incorporate both in a single doesnt really make sense.. goes against the grain of what each style is supposed to do..
  8. mahras2


    Going against the grain is exactly how to reap profits. Every average Joe Shmoe knows about mean reversion and trend following. Creating an unorthodox style is the method of making money in a zero sum arena.

    I am sorry but I really cannot give any more information out without disclosing the parameters or the edge of my system.

  9. I love people who talk about some magical system that is making them money.. the truth is that anything u thought of has already been tested many times..

    I would love to see the backtested results of the edge you have..
  10. Its going against the grain of logic... when a market is trending is not reverting back to its mean..

    What you are doing is basing a thoery of an inverse relationship.. which has no true edge.. bottom line.. either discuss your system or show us backtested results.. otherwise your posts have no value to this community...
    #10     Oct 9, 2005