Will a partial martingale system work?

Discussion in 'Trading' started by newguy05, Feb 19, 2008.

  1. Feel free to state why it doesnt work, but lets keep the flames down pls.


    1) System has a 60% win/loss ratio for each trade

    2) All trade placed has a stop loss at -3 tick, and limit at +3 tick for exit. In other words, reward=3tick, risk=3tick (yes 3 tick is picked arbitrarilly, obviously the # depends on your system, but the key is stop loss = limit )

    3) Instrument traded is something very liquid, such as the ES

    Now we all know martingale is not a good strategy if followed completely, as it exponentially increases your risk as T approaches infinity and P approachs 1. But what if you just follow it for the first x trades to a point where your max loss is reached, then start over?

    For example:

    --------- 1 UNIT-------------------
    1) 1 contract
    2) 2 contract
    3) 4 contract

    Total risk: 7 contract
    Win: 0.6 Loss: 04
    Probability: 0.4 ^ 3 = 0.064 -> 6.4% loss + 93.6% win

    1 UNIT x 100:

    6.4 * (7 contract) = 44.8 contract loss
    93.6 * (1 contract) = 93.6 contract win


    If not using above, normal trading with 60% win/loss system will yield:

    44.8 contract loss
    ( 44.8 / 4 ) * 6 = 67.2 contract win

    as oppose the partial martingale approach of
    44.8 contract loss
    93.6 contract win

  2. It sure looks good on paper.

    however, with all the unknown variables that forms todays markets, youre likely to need more than that to squeeze cash out of the market with this strategy.

    also, martingale only has a 100% chance of succeeding if you have unlimited funds

    all you need is a couple of losses in a row, and no matter how high your win % is, its bound to happen, can your account take a hit like that? especially at the beginning?
  3. how do you get a 60% winning %?
  4. I have been "playing" some trend following types of approach where you can do a controlled martingale if you have markets that are good trenders, even if they are merely swings. The key is to start reasonably small and then increase size controlled and only to a certian maximum.

    For example, imagine you get a long signal at end of day and go long 5 ES contracts. 2 days later it flips to sell and you cover and go short 10 contracts. 3 days later it flips to buy again and you go long 15 contracts.

    Some systems have an average of 2 - 3 flips during choppy consolidation periods and then find a decent move out of the chop. When the chop ends there is usually a significant move that you catch with larger contracts. Now this is more swing than pure trend as ES might breakout and move 40 pointson the last move. So you catch the move out of the chop with bigger size but you still have to limit yourself to no more than 3 escalations.

    I am computer bad so backtesting is slow but I have noticed that if you have good trend following/swing signals, the chops are about 2 - 3 flips. Of course the past year has seen some great swings so it may look good now.

    Martingale on its face is not really recommended but using the analysis of scaling in on winners, the idea is if you have results in the system that you get a significant move on breakout of the chop that you can set rules where you scale up on certain types of chop.

    More theoretical now but playing with it to see.....
  5. What does a "60% win/loss ratio for each trade" mean?

    Beats the S&P out of me