How much to bet when you have an edge

Discussion in 'Risk Management' started by jacksmith, Feb 27, 2009.

  1. We had come out a system. Backtest shows that the system can predict the direction of market movement correctly in 60% of time. (It is a system to predict if market will go up or down, but not how much). We believe that 60% correct predict ratio shall maintain in future.

    Given these information,
    1. What percentage of capital shall we bet each time ?
    2. How to set stop-loss ?
    3. How much leverage shall we use ?

  2. oraclewizard77

    oraclewizard77 Moderator

    1) 5% - 10%
    2) Set stop as a % of loss, for example if you are risking 10% position size, set the stop to take the loss after you lose 10% of your portfolio.
    3) No leverage should be used.

  3. If you expect to win 60% of the time and you bet 10% everytime, you should expect HUGE drawdowns.

    If you play this 10000 times the chances of losing 10 times in a row are near certain. Because its a "fixed fractional" bet your equity won't go to zero, but 70 to 80% would be expected. Once you take a hit like that, its near impossible to get back to even.


    to OP: you aren't providing enough info to get a good response
  4. oraclewizard77

    oraclewizard77 Moderator

    I should have said not to compound returns. For example, lets say you start at $ 100,000 and you make $ 10,000. Your next "bet" would not be 5% of $ 110,000, it would be 5% of the original $ 100,000. If you are trying to grow returns more quickly then you do have to risk a higher % of your return and will therefore larger draw downs. Obviously, the larger account size you have, the less you need to risk.
  5. Think about it the other way, if you lose the first bet, the 2nd bet you would be better 10K with 90K in equity. Betting a fixed amount means you go down faster and up slower than if you bet a fixed percentage. Unless you are coming out of a draw-down or coming off a peak.

  6. Mr J

    Mr J

    If we know we'll hit 60% going forward, 10% is fine. The fluctuations of bankroll will be massive, but in terms of risk of overbetting it is fine according to the Kelly Criterion. The problem is that we don't know we'll hit 60% going forward, which is why we'd bet a fraction of Kelly. Probably no more than 5%, which I will add will lead to huge fluctuations anyway.

    Agree that no leverage should be used. Leverage is completely misunderstood by almost everyone. The only reason for it to be used is to allow for diversification, such as keeping only a fraction of your capital in your trading account or allowing simultaneous trades.

    This varies depending on strategy, situation and personal comfort.
  7. No system can "predict" anything. Systems can have only some varying rate of success. Prediction means nothing in trading. Only astrologers claim they can predict the future. You can be correct about direction and take a loss at the same time due to volatility swings.

    How in the world do you know your system will be correct 60% of the time if you don't know your stop-loss?

    Give me a break people. Nobody noticed that?
  8. There is a thing called the risk of ruin, it's a formula. You may want to use it to dertermine how much to bet. And yes, you did not give enough info for us to answer. In fact there is no one answer, it depends how much risk you personnaly want to take.
  9. I noticed.

    With all due respects to the OP, he sounds like an amatuer systems developer.
  10. A very good point.
    #10     Mar 1, 2009