Risk $1000 to make Millions

Discussion in 'Risk Management' started by VoodooMMI, Apr 30, 2008.

  1. bg514

    bg514

    Here's an idea: run two systems at the same time: one that only goes long, and one that only goes short. If both systems call for identical opposing trades you do nothing. These bets can be "imaginary" until one of the systems gains the upper hand. for example, if the short system says sell 3 and the buy system says buy 5, you bet the difference (2 contracts) on long. if either side wins at the 8 contract level then the winner will have won more than the loser will have lost, and at this point the loser retires. It now becomes a one horse race with no hard and fast rules.
    An anti-martingale system, taken to the bitter end, always loses it's biggest bet. The question then becomes when to pull back and if so, how much? This is where knowing how to trade would come in handy. For myself, if I could run this progression up to the 55 contract level and win, I would probably shut it down. There's a saying on Wall Street: Trees don't grow to heaven!
     
    #31     May 15, 2008
  2. Bowgett

    Bowgett

    #32     May 16, 2008
  3. Voodoo:

    I read the comments in response to your post, and I could not believe that most posts seem to have mis-understood your idea. The central idea in your post is that at earlier price levels you shoulder losses with your cash, and beyond a threshold, the losses will be taken care of with the accumulated gains. The threshold level is that point where if you look below you have a loss (which is fixed and known), and if you look above you will see an increasing amount of gains. I have various ideas related to this, but let me start with a few:

    1. What you just did is a essentially a call synthetic (but exotic) option (let's call a Vooddo Option (not in the negative way, just to differentiate it from other options)).

    2. One can design other Voodoo options, and your design falls under a larger class of an-martingale. Let as assume that at a given point in the price axis you add a new size denoted X(K), where K is the new level. Your new position size is Y(k)=Y(k-1)+X(k). Let us call G(k) the accumulated gains at step K. K* is the level where you get stopped with no loss and no gain (but there exit design where there is no such K* as all what is needed is that above K* there is gain, and below it a fixed loss). K* can be thought of as the strike level of your Voodoo call. If Step is your increment, your system is a member of a larger class of Voodoo systems:

    Case 1: When K is greater than the threshold: Y(K)*Step <= f(k)*G(k-1), where f(k) is a design variable which is non-negative but never higher than 1. If stoped out, then your gain will be (1-f(k))*G(k-1).

    Case 2: If K is less than the threshold: Then Y(k)*Step=FixedCapitalAmount. The initialization is then more specific than what happens after the threshold.

    3. To analyze your system and any system, in that class you need to determine K*, and then calculate the probability that your system will end in Case 1 and in Case 2.

    4. To answer 3, you can take various routes:

    4.1 Simulation
    4.2 Direct maths
    4.3 Transform your system to a set of exotic equivalent options (barrier, touch, binaries, etc..)
    4.4 Analyze it recursively, and then solve it using a recurrence equation solver, or by induction.

    5. The most important point I think you have to re-look at is your entry. You use price, but you seem to have forgotten a crucial element in this: time! Because if you enter independent on the time element attached to price, then you might be taken out more frequently, as the probability of touching a price is always higher than the probability of ending up above/below a given price by the end of a certain time, and also because amplitude of cycles may play in your favor or completely against you depending in which cycle you are in.

    6. Also note that there are other design possible where you can change Step (size, linear vs. log, change as a function of K, etc).

    7. Point 6. also applies to f(K) is the above.
     
    #33     May 24, 2008
  4. thn5625

    thn5625

    Good post risktrader. To the thread starter: have you considered that your method can be used if you trade once or twice a year? What I mean is that ideal condition in which the price moves smoothly down without retracement doesnt happen often. (and this is very very ideal and not necesarily realistic) Moreover, it is very difficult to "know" when it will happen. I hope you do not think this is a daily strategy for money mgt b/c the probabilities of getting that straight down market movement is rare.

    My suggestion is to use when you are in a "black swan" event. I am not joking here but you have to limit this strategy to only events such as that. Again, the reason is b/c you want a smooth, incremental, down movement of your stock/index and this usually happens when most people are not expecting it and are caught on the other sides with there pants down.

    Also, if you think about it, dont discard this idea. Instead, relax the aggressiveness of the positions so you can use it on a daily strategy in which you can trade more often. You will just have to be happy will 10K return vs 2.5M :)

    One more thing: Make your strategy more dynamic. It is too "fixed" to actually work in the markets. Your thought exercises in trying to make this method viable probably stimulates your intellect but "stimulating" your intellect will not translate into the returns that you are hoping for. Dont make trading harder than it is --> a dynamic strategy that changes according to what the market gives you is a better bet. My suggestion is to trade a much much more conservative version of your strategy and once you understand the price action. Until you understand price movements, you will not know whether or not your antimartingale method will actually work as planned.
     
    #34     May 25, 2008
  5. not a chance
     
    #35     May 25, 2008
  6. re-read what this person has written, what he tells you is solid good advice. Here are some calculations that support his point:

    If p is an approximation of the prob for price to step back by 20 points, then your probability to be in case 1 (never higher than the breakeven) is: (1-p)^(k*).

    People may think that p is static and/or less than 0.5, but it is not. Why? Because it depends on your holding time of the position (and of course the size of the step). If price moves up by 20 points quick, and you hold time is say T, then the probability for the price to go back down by 20 points is actually TWICE the probability of the price ending below your 20 points down at the end of T. If the letter prob is around 40% (less than 50%), then p is in the area of 80%.

    Now (1-0.8)^6 is a small number. It represents a small prob for you to be in the positive area.

    One improvement you can make is that if price moves suddently by 20 points, you may wants to hold 20% of time before adding, as the average time for price to move done by 20 points is around T/5 (the distribution of stoppage time is skewed to the left).

    The above probs will require you to use a tiny fraction of your payroll for this. (may be 1/50 to 1/100th) in each bet.
     
    #36     May 25, 2008
  7. Spoken by the multi-alias Stalion - RiskFreeTrading. The "veteran" unmasked as a Eunuch LOL

    [​IMG]
     
    #37     May 25, 2008
  8. asap

    asap

    an anti martingale strategy along the lines of what the op describes would yield a profit factor of north 2.0 with a sterling ratio in the range of 3.0 if used during the last decade (which was very tough on always long strategies).

    the op has hit the nail on the head imo. the essence of intelligent trading (let your winner run, cut your loses soon) coupled with the edge of he index positive drift makes an exceptional trading expectancy. however, one should accept several losing years in a row are a possible and likely scenario. however, after any of those multi year dd's, the system rapidly build into a single winner to achieve a new peak in equity.

    see in attach the equity curve for oneof my dax ats based on these assumptions, trading one lot (adding up to 10), just doing longs and covering the bear market of 2000. the system made 300k while the underlying is below the initial level.
     
    #38     May 25, 2008
  9. asap:

    Could you pls explain your system with a small example, and also explain the numbers and curves in the file you attached. It is intriguing, and I want to learn more from it. Thank you.
     
    #39     May 25, 2008
  10. voodoo,

    I am a total newbie but I have a system that might give you better results..

    Send me a check for $1000 and forget your betting idea's and you will thank me..

    You must like Black Jack as I have seen some guys play that way..

    I sincerely wish you luck and hope you prove everyone wrong..

    All the best,


    singlefather


    ..
     
    #40     May 27, 2008