Pyramid positions

Discussion in 'Risk Management' started by Aceoface, Jan 26, 2013.

  1. Aceoface


    I made countless simulations of a version of the turtles system with pyramiding positions but could not get better equity curve statistics than the non pyramiding version. When pyramiding with a risk of say, 0.90% per ATR, i could get the same CAGR% with better stats (better sharpe, info ratio, MAR, less trades) with 1.3% risk per ATR and no-pyramid. The risk per trade is the same because pyramiding means a higher number of ATRs at risk. Yet non pyramiding looks better.
    Anyone has a contrarian view on this?
  2. Sergio77


    I do. Avoid pyramids.
  3. Aceoface


    Its not that pyramiding hasnt got advantages. CAGR% increases substantially. However there is a catch (more than comensurate volatile equity curve, lower MAR, lower sharpe etc). I wonder if there is a clever way to pyramid and improve these stats as CAGR% increases.
  4. Pyramidding requires substantially more capital than it's worth to worry about doing single entries and exits.
  5. horton


    In my view a proper pyramid is built using multiple valid setups in the same instrument whereby the risk of each setup is managed independently. You have to set a maximum position size for the instrument to acknowledge the risk of a fast market or gap event, and you do not take a setup if the prior setup has not reached a break even posture.

    Today for example I took two valid setups in the TF in the manner described above such that I never risked more that 5 to 10 ticks and exited with about 35 ticks. Actually along the way I had two scratches that cost me 2 ticks each.

    So this is working for me but I would appreciate feedback.
  6. Daring


    I add to winners and to losers but using different techniques for each.

    At the same time Im not against pyramiding or averaging down or in favor or against it.

    Trade and let trade.
  7. Thanks for sharing information. Actually I had also the same question in mind for a long

    time anyways you started this thread & I am so happy.
  8. First match the data accurately. Then change one thing at a time and you will have solved your own problem.