Position sizing

Discussion in 'Professional Trading' started by sheepsucker, Oct 25, 2009.

  1. I trade futures and have trouble determining a good model for position sizing.

    I have read quite a lot on the subject and ran lots of simulations and am aware of the theory in the field. Kelly, Monte Carlo simulations etc.

    My intention is to swing for the hills like neke.

    When I am trading well I have I have RR 5 and win-% 35.
    A working approach seem to be to risk 1-2% per trade, at least Monte Carlo simulations look beatiful.

    If I get out of touch somehow, which happens unfortunately. I am quite quickly, after 6 losers down ca.10%.
    Which can affect my trading so I get down even more.

    So I have been thinking after 3 losers cut down to 1% and after 3 more losers to cut down more and even hit SIM.

    Another approach I have been pondering would be using levels. example:
    -start with 20k and risk max 1%
    -when its at 22k increase bets to 2-3%
    -if again down to 20k get back to max 1% bets

    What im really thinking about is some rules to use to maximize potential but still guard for psych-issues and varying market conditions.

    Thanks for any help!
  2. bone

    bone ET Sponsor

    1. Start out trading one lots or two lots. Your only goal for the day should be net positive.

    2. Two net positive days in a row, up your size one lot.

    3. Two negative days in a row, go back to your previous size.

    4. If you started out trading one lots, and you can manage to average two bumps per month, you'll be a 25-lot trader in a year.

    That simple - please don't make it more complicated than it needs to be. If you want to start out trading 10-lots and then go with two-lot bumps, that's fine if you're capitalized correctly.

    A Monte-Carlo simulation for position sizing makes the false assumption that your trading system's edge is constant. The simple system I described above is a performance-based system.
  3. No disrespect but I do not agree with your recommendations as they might be abit too simplified.

    If you trade fix lot size you are not position sizing. Risk must be kept constant each trade else you risk losing big and winning small.

    I agree with you for upping size with good performance.

  4. Fixed Fractional is the simplest and most reliable. You basically make your total trades as a % of your trading capital.

    Stay away from Kelly and optimal-f

    You CAN attempt to reduce risk by mimicking some algorithms, which reduce exposure when you are in a losing period.

    Fixed ratio is another popular one, but from what I have seen, fixed fractional tends to make more sense and has fewer oddities.

    You can do a google search for details.
  5. What are some of the algos for reducing exposure in a losing period? Pls share.

  6. Hi,

    Thanks to everyone for the advice.

    Most suitable for me currently looks a combination of Bones down-to-earth approach and fixed fraction. That would allow for good optimization but still benefit from the simplicity of bones approach with managing the ups and downs of performance and edge.

    Start with for example risking 0,2%/trade.
    2 Days green, then 0,4%.
    2 Days red, then 0,2%.
    4 Days green, then 0,6%.
  7. They basically boil down to: when you are in a downtrend, you reduce your positions even faster than you grow them. Not rocket science, for example: when you are increasing your account size, you might increase your positions at x%. But if your account is shrinking, you might reduce your positions at 1.5x% or whatever.

    The percentages are totally up to you - there is no magic forumula here, regardless of what anyone prints or says. You need to do your own backtesting or sim or practice trading, and find out generally works for YOUR system and YOUR tolerance for pain. Remember, protecting your capital comes before making profits. If you sit down and chart out how to earn 1,000% in 9 months, you will learn it does not work like that, and the higher your leverage, the more likely you will exceed blowout (think, beyond bankruptcy)...

    Van Tharp also did some position sizing work (I believe it was volatility based), but I never personally read it.

    Position sizing algorithms in no way exempt you from risk-adjusted reward realities. It is mostly an exercise in trying to balance risk and reward.

    The smartest thing I have seen, is to not change your leverage as your grow. And the more you have, the more you should DEcrease your leverage.

    For what its worth, I once saw a paper that said that tested many, and found a leverage factor of about 7-fold seemed to be an optimal tradeoff between growth and risk (this was for futures trading only). This ignores volatility or market conditions, which greatly affect this concept.

    And of course, this post in no way makes me or anyone else here responsible for your losses. Like other forum posts, you always use everything at your own risk, peril and detriment. Research things before shooting your own foot... :D
  8. Stok


    bone...what if u r trading 20 lots, or whatever number bigger than just one or a few...is there a certain percentage bump you use as a rule? For example, of one was trading 20 lots, with 2 positive days, would u up contracts 1,2 or what number based on the original base of 20?
  9. Nothing's too simple.
  10. Yup.

    Another idea is a certain volatility based stop (ex. mult. x ATR) used with a fixed $ loss for each trade. This will position you for volatility too.
    #10     Oct 26, 2009