Discussion in 'Risk Management' started by macrotrader, Mar 25, 2012.

  1. Say I think a stock will go from 100$ to 50$ within 2-3 months. Also, I believe that if it retreats it is unlikely to bounce. What I mean is, if it goes from 100$ to 70$ I think it is very unlikely that it will go back up to 100$. This is because I believe that people will eventually see that the stock should really trade at 50$ or lower, because the company doesn't have any substance and is only temporarily hyped (or even promoted).

    Now I could simply put on a standard position, say 5% of my capital. However, I want to use eventual profits to increase the position, although my initial risk should be no greater than 5%. I want to pyramid 4-5 times to get up to 20-25% for example. I could calibrate this by using volatility.

    I recently did a trade where I bought a stock at 6$ and doubled up after it went to 8$. I sold out at 12$, but in hindsight I should have be more aggressive when things were going the right way.

    Any thoughts?
  2. I never understood pyramiding because the volatility of the retraces usually kick you out. Stuff usually does not go up in a perfectly created uptrend or down in a perfectely created downtrend.

    Why not increase size from the get go by maintaining your risk at the same levels but obtaining more leverage by using options?
  3. yes, Martingale already figured this deal out. It always works if you have an infinite supply of money.

    Gernerally pyramiding always uses 4 units. If it moves against you you add 5 then 6 then 7.

    If it moves for you you add 3 then 2 then 1 and that is a full load.

    It's a beautiful thing when the whole pyramid moves your way.

    But still you need to guess. And any math man will tell you it would have worked out just the same if you had guessed right in the first place.

    So pyramiding is just a psycho technique to help you find your risk tolerance.

    I do it all the time, but I would have the same results (or maybe better) if I just put it on and used a hard stop.

    only thing I can add is introducing time instead of just price

    (I'm not just old, time is also very important to me)
  4. Brass


    Yes. I think that if you believe you can predict price levels and, even more to the point, within a certain time frame, as well as price movement personality along the way, then you are dreaming.
  5. Brass


    Pyramiding does not normally imply Martingale. Pyramiding is normally undertaken as price moves in the direction of your trade.
  6. Pyramiding is only possible for a strategy that can be pyramidded.

    In the case of two strategies trading the same instrument when they're in synch is when you can and should add additional units.

    I call this either and extra, but it'll only work if you have a strategy that's relatively in synch with the other.

    I place 2 contracts on NQ when I have a pairs signal, and 1 contract when I have an ergodic signal. In the case where I have both I doubled my profits while adding 1 basis point to my drawdown by adding 2 more whenever they are in synch, so that I have 5 if they are all in agreement. Sometimes I will have the ergodic reverse direction, in which case, I close the extra and leave my position whichever direction the pairs signal is pointing, leaving 1 contract open, after closing 4 lots of the 5. I will only leave additional leverage open while they are in synch, so as soon as they are not, I revert back to my normal position size then leave the net +/-2, -/+1 for net 1/-1 position without agreement. When they're in synch I will have 5, and wait for one of my signals to close, leaving the net of the position open, usually 2-1 or -2+1 if the ergodic reverses.

    Mine is a unique situation and unless you have strategies where you can do this, it's best to think more about the initial entry than worrying about when to add. If you have strategies designed to pyramid, it's a piece of cake, but adding based on discretion will inevitably cause the martingale blowout.

    Pyramidding is fine if it can be automated, but when there's discretion the tail risk becomes way too large to even consider that as an option for trading in the long term.
  7. If the stock follows a random walk in the sense that my scenario doesn't change, pyramiding doesn't make sense, because I can simply risk my initial capital and leave the position unchanged. However, if the probability of an upward move goes down it makes sense to add to the position. As I understand martingale means adding to the position if it goes against you (because the probability of a reversal increases). No infinite wealth required for my approach. I am not sure, but I think Soros reflexivity theory refers to the evolution of the market. As soon as the perception changes and a structural break occurs, a reversal is very unlikely. Which means, if you stand on the right side, you should increase your position, and if you're on the wrong side get out. The rule let your winners run and cut your losses short only makes sense if one makes assumptions about conditional probabilities. I did a gazillion backtests and not ounce tested pyramiding, because for a statistical trader it doesn't make sense IMO. Same is true for stop-losses, a concept which seems to confuse a lot of people.
  8. Brass


    Everything is more clear after the fact. But when you're testing the depth of water in real time, you should probably not do so with both feet.
  9. ??? We do it all the time in all aspects of life. Just going to the grocery store, or buying a new car, or a computer, you have an idea if price is full or will soon be reduced.

    I agree, difficult to make a living guessing what price will do without some sound money management. But still the guessing part is for me an important component. Not a big component, but neccessary to turn it from just another breakeven year (minus the spread and commissions) and something which will give you a tax problem.
  10. Brass


    If you are relying on prediction for your trades, then I suggest you wear your lucky charms. The only trading that I personally know to work involves reacting to events (price action) that you can actually see developing in the present. Compare that to knowing where prices will turn, by how much and when.
    #10     Mar 25, 2012