Acceptable Max DD?

Discussion in 'Strategy Building' started by Corso482, Jan 22, 2003.

  1. DT-waw


    acrary: you're right. It wasn't a good example. I just wanted to say, that if two strategies have the same return/risk ratios and one has higher risk than the other (but also higher return) - choice depends on your risk tolerance.
    #21     Jan 24, 2003
  2. acrary


    I understand. I was just pointing out that the risk tolerance can be adjusted using money management. That way all you need to know is the annual return / max. dd to make an objective decision. If the first example had a 50% return and a 30% max dd, but my risk tolerance was only 15%, then by trading at 1/2 size, the return drops to 25% while the risk declines to 15% and is superior to the 15/15 strategy. If both strategies had the same reward:risk ratio, I'd pick the one that trades the most often (to take advantage of more compounding opportunities).
    #22     Jan 24, 2003

  3. i agree w/ surfer, this is not an accurate generalization.

    it's true that the more times you flip a coin, the more statistical aberrations you will see- i.e. the longer you're at it, the more freak streaks there are of heads and tails. i'm assuming this basic notion is where you get your assumption from, that eventually you will hit the mother of all loss streaks through pure chance as the outliers expand.


    1) a rational and robust trading strategy applied to markets dominated by a constant set of factors (fear, greed, structural continuity etc) is not the same as a progressive series of coin flips, not by a long shot. Statistical aberrations are far more likely to be bounded in the real world when intelligent decision and analysis is involved, especially if your method is built on embedded market principles that have not changed. Actual infinites are far more rare than potential infinites, and the two are often confused.

    2) if you have a rational and conservative strategy and a reasonable winning percentage, it would take longer than you might think for the killer losing streak to materialize. if you were risking a percentage of initial capital per trade, the length of time would be even longer/approaching infinity, given that your capital at risk decreases as your capital base decreases and thus your risk asymptotes out above zero. The probability percentages of total ruin rapidly approach irrelevancy given enough starting capital and a conservative enough approach. At some point- perhaps the point where the percentage odds of the freak streak are on par with getting hit by a piece of Skylab- the notion becomes pointless.

    3) going back to the coin flip example, there is no reason why you would necessarily have to go bankrupt even if the aberrations were completely unbounded and you had an infinite length of time to play. because there is no average distribution for the outliers (they can only get bigger as time goes by), there is no reason why heads couldn't get a jumpstart on tails- akin to wins getting a headstart on losses- and never give up that lead. Win streaks could get out in front and stay that way forever, with the two getting farther and farther apart as the outliers keep expanding. (This sort of thing happens to people in life- thus the expression "some people have all the luck".)

    (note that there is always the possibility of your strategy becoming invalidated, or the blowup that takes you out of the game in one day, but I'm assuming those occurrences were not a part of your basic proposition).
    #23     Jan 24, 2003

  4. this generalization isn't true either; it completely discounts the edge of experience, patience and foresight. take a guy like Jim Rogers for example. And the "much better" notion is blown to smithereens if you compare in absolute dollar returns as opposed to percentage of capital ones.
    #24     Jan 24, 2003
  5. well said, dark. thank you for elaborating on my initial reaction to man's post.


    #25     Jan 24, 2003
  6. One of the systems I trade had a bad drawndown in the past, but it is now doing extremely well, in the last 12 months only one was a small loser. I guess that I was lucky that I did not know about this drawdown because it happened in 2001 and I had backtested the system for 2002 only and had started using it because I liked its simplicity. My point is this: had I known about this drawndown I would now be losing money by not making it using this system. So sometimes ignorance can be bliss and too much information can be a problem.
    #26     Jan 24, 2003
  7. onelot


    hi acrary,

    how often did they calculate that?
    #27     Jan 25, 2003
  8. acrary


    Our positions were marked to market intraday. Evaluation was done yearly, however if you were clearly out of bounds, you'd be gone long before the year was up.
    #28     Jan 25, 2003
  9. onelot


    ok... i thought you were gonna say you had to do it weekly... even monthly would've impressed the hell out of me... :)

    not that yearly doesn't :)
    #29     Jan 26, 2003
  10. huuuhhh... man... man... man....

    I agree with surf and darkhorse...

    It sounds like you're talking from a 2nd or 3rd hand information. Time to get into real research.
    #30     Jan 26, 2003