Position sizing /cash allocation for intra algo portfolio

Discussion in 'Automated Trading' started by Daniel.a, Aug 27, 2019.

Algo Portf, allowed max DD

  1. 5%

  2. 10%

  3. 15%

  4. 20%

  5. 25%

  6. 30%

  7. 35%

  8. Kamikaze + %

Results are only viewable after voting.
  1. Daniel.a

    Daniel.a

    Hi all, I have a a portfolio containing a multiple of strategies on 5 markets..
    NG,CL, ES and S all intraday strategies, and then i have one strategy mean reversion sp500 stocks 1-5 days max hold.

    I cant decide on a portfolio funds allocation strategy, i am locked to a known stop per trade, and i am locked to a constant position size while in a trade.

    Each market contains 4 strategies, and can all go long or short.

    Question,
    So am leaning towards a simple allocation strategy, by looking at portfolio max Drawdown total or max Drawdown intraday for the past 20 years and then allocate capital enough so that max drawdown represent the wanted max allowed drawdown of my portfolio for example 15%...

    Or a capital allocation based on Montecarlo of the whole portfolio, how would you model this? look at what what 5% probability max dd represent of a montecarlo run of porfolio, and use that DD as a number and allocate capital enough to have that number represent for example 15% of the portfolio capital?

    Idea or suggestions ?
     
    Nobert likes this.
  2. I think @globalarbtrader's book covers this in a way that would probably appeal to you.
     
  3. I agree with this advice. I haven't read his second book, but his first book ("Systematic Trading", by Robert Carver) also touches on this.

    I'm not sure whether the account's DD is the right parameter to focus on, as there are here multiple strategies at play, each having their own time scale. And these strategies could be positively or negatively correlated with each other. Maybe it would be better to assign a certain amount of maximum risk to each of these strategies.
     
  4. Daniel.a

    Daniel.a

    Thank you @nooby_mcnoob and @HobbyTrading, i hav Roberts book and read it a year back i think, might need to revisit it then.. and yes i agree that looking at max DD might not be optimal or the best best thing would off course be to subgroup some markets and risk to a subgroup correlating the most etc... however i am kind of locked in to what i can do in my current structure of entry/esit strategies and mm strategy.... so i am locked to having stop as a fixed dollar amount and i can limit positions as a total to whole portfolio but but into subgroups..

    So i am looking at a simple and effective way to allocate capital, i was looking at max DD, or intra DD, and then allocate capital and adjust to get the risk i want..

    I was just looking to get ideas of how others would do it, and i was also interested to see if anyone was working with MonteCarlo and DD together to decide capital allocation...

    Anyway i will read the book as suggested again, thanks
     
  5. Max DD is a poor choice to optimise because for a given backtest its pretty random what the DD is (to be technical it has high sample uncertainty)

    GAT
    AKA Rob
     
  6. Daniel.a

    Daniel.a

    Hi Rob, Thank you for your input. You also believe this is the case if you use Max DD derived from a Monte Carlo test?
     
  7. Something like median of the MC distribution of maximum drawdowns?

    That would be much better although I haven't formally investigated the statistical properties of MC max drawdown.

    GAT
     
  8. Daniel.a

    Daniel.a

    @globalarbtrader
    Yes, for example 50 runs of MC for your diversified portfolio, and a median of those Maximum drawdown at a 5% probability, and then you use that amount x2 as capital allocation for your portfolio.

    Regards
    Daniel
     
  9. Out of curiosity, what is his book called?
     
  10. You can find the details here: https://www.systematicmoney.org/
    The "Systematic Trading" book describes, among others, how to handle risk management.
     
    #10     Sep 7, 2019
    Snuskpelle likes this.