Acrary is a genius!

Discussion in 'Strategy Building' started by greaterreturn, May 4, 2008.

Thread Status:
Not open for further replies.
  1. Unless I'm missing something, all you need is a spreadsheet program to add each data point of the two non-correlated graphs.

    There's a free one called open office on the web.
    ----------------------------------------------

    P.S. I was looking forward to hearing more interaction from maestro and acrary.
    I've looked over some of acrary's stuff, and it looks pretty interesting.

    Too bad the hecklers have to jump in so quickly. It's rare to find useful discussion here, why ruin it and then complain later that ET is a wasteland of
    useless forums:confused:

    P.S.2 Maestro, I always thought cybernetics was some sort of quack science (thanks to holistic mind training associations made in the 70s). However, after looking at some of ashby's stuff (and considering the collaborative efforts of the likes of von neumann and c. shannon), I've gained a bit more respect for the science. Thanks for mentioning Ashby.
     
    #101     May 8, 2008
  2. Very insightful point (that probably nobody has mentioned before), and I would have to agree with the common view that Only geniuses would realise and understand the new discovery that finding more uncorrelated profitable strategies is a holy grail in trading!

    A newbie like me would have hardly even found just One --- not holy grail but profitable strategy!
    :confused:
     
    #102     May 8, 2008
  3. man

    man

    i think for edge testing and other alan-like-studies excel
    should do the job. most of his chartwork finally took
    place in tradestation.
    and afaik tradesignal allows combination of various
    strategies.
     
    #103     May 8, 2008
  4. man

    man

    i call that stresscorrelation. you take two equity curves
    A and B. now calc their daily change, use logreturns, not
    percentages. now just take those days where A was
    down, this is timeseries 1. now for each day in timeseries1
    you take the dailychange value of B. that is timeseries2.
    now calc the correlation between them. doing the same
    the other way round, starting with B, you end up with
    two stress correlation figures. take the mean of them.
    done.
     
    #104     May 8, 2008
  5. man

    man

    seems like you did your best already ... providing a paper.
    i guess we just have a semantic problem here. in the
    years on et i sensed that people who came from discretionary
    trading have a specific way to look at things. hit ratio,
    payoff ratio, profit factor are very traderoriented figures.
    whereas sharpe, sortino, sterling, treynor, whateverhaveyou
    ratios are not the thing traders can connect to. my
    background is investing in hedge funds. there you do
    exclusively think in percentage terms and ratios of the
    kind mentioned.

    take care.
     
    #105     May 8, 2008
  6. man

    man

    tough to figure out the route your sarcasm takes in
    this post of yours ... :) ...

    but there is something very important embedded: finding
    uncorrelated stuff is very difficult to do and to organise.
    it is so difficult to really figure out working, unfitted
    strategies that developers find it hard to do it all over
    again at some other (uncorrelated) place. most of the
    time they end their new development by finding out
    that they are going to trade the very same effect as in
    the last system, just coming from a different angle.

    this is actually one of the most important function boards
    like et provide: additional views. other perspectives.
    for example this thread. i am 95% sure that maestro's
    points do not add anything to my current setting. but
    the remaining 5% and his reputation as a reasonable
    pro will make me check out the sources he mentioned.
    and once in a while such thing takes you forward.
     
    #106     May 8, 2008
  7. I thought I was very seriously serious! :D


    Probably that's why, when and where a black swan would kill us!
    :mad:

    Bye! :)
     
    #107     May 8, 2008
  8. minmike

    minmike

    What I have found is combination of combining starts helping and hurting. Combeing doesn't provide as much synergy as I would have thought.

    strat A has a 1 sharpe. strat B has a 1 sharpe. completely uncorrelated might get a 1.1 or a 1.2 when combined. I haven't found much more improvement than that.

    But if strat A sharpe 1. Strat b sharpe .5.

    It has an averaging effect that seems hard to overcome. combined 1 to 1 I would think the sharpe would come out in the .75 to .9 range. Just what I have experienced, rough estimates.

    Great thread. you guys helped me today.
     
    #108     May 8, 2008
  9. man

    man

    no easy decoding your post ...

    adding similar systems together leads exactly to what
    you describe. once you have systems that might be
    negatively correlated, the picture changes substantially.
    but in principle you are right: just throwing in different
    strategies ain't enough. two ways to increase sharpe:
    higher frequency and non correlated stuff.

    nevertheless, i would estimate it takes ten times the
    effort to turn a sharpe 1 into a sharpe 2 and another
    twenty times to move from 2 to 3. i am currently in
    between 1 and 2, so my talk about 2 to 3 is theoretical.

    and one needs to row twice as hard to remain at 2 than
    to stay at 1. quant trading is an uphill battle.
     
    #109     May 8, 2008
  10. minmike

    minmike

    I think you got most of what I was saying. i talk much better than I write.

    I have only found non correlated systems. I have yet to find two profitable negatively correlated systems. Have you had better luck with negatively correlated systems than with non correlated systems? I have always looked for non correlated systems (partially because they seem easier to come up with.) Maybe I should start looking for negativle correlated.
     
    #110     May 8, 2008
Thread Status:
Not open for further replies.