What do you base your % risk on?

Discussion in 'Strategy Development' started by kevmay21, Dec 11, 2006.

  1. kevmay21


    I have ready in multiple books and online that 2% of capital risk per trade is a good acceptable leve. Once you start going over say 3% your pushing it.

    What I am wondering is what is this based off of? Is it the largest loosing trade if your trading a basket of markets?

    For example I have backtested a system on Lumber and the Yen over 26 years that has net profit of 302k, Max drawdown of 25k, largest loosing of 5,575,, average loosing trade of 744, 42.13% profitable.

    So in this example with a Largest Loosing trade of 5,500 you would nead over 200k just to trade a basket of 2 markets with a reasonable max drawdown.

    Or is it better to base it off of aveage loss if you standerd deviation is reasonable?

    Seems to me it would be hard to trade a system with a starting balance of say 40k if your max loss per trade could only be $800. I mean most systems for them to perform properly it seems you need to let them follow there natural flow and not set hard $ stops.

    Any insights or suggestions would be greatly apreciated. Not looking for handouts and have been doing Lots of reading on the subject.

    Thank you

  2. I use a risk level that allows me to sleep at night and avoid gut sickness. There is no way to know the length or magnitude of the next losing streak. Back testing only describes the past.

    I examine the equity log of the portfolio simulation and ask myself if I am comfortable with the losses.

    I have a uncle point. If I lose 50 % of equity then I stop trading for a year. The last time I remember a 50 % loss was about year 1991 when I did not stop a loss on a short AMGN position.

    The best teacher I know of is a big, painful, ugly, expensive mistake.
  3. Some of my systems use % heat and some % risk. My moving average crossover system uses % heat. A 1 % risk system uses a hard stop and attempts to limit risk at 1 % of equity. A 1 % heat system can lose more than 1 % of equity. In the case of my moving average crossover system the exit signal is generated when the value of the fast moving average becomes less that the value of the slow moving average. If price is decreasing rapidly then the loss can be much greater than 1 % of equity.

    I think it helps to have a portfolio of more than 10 instruments because the draw downs tend to be of shorter duration and of lesser magnitude. The portfolio effect reduces volatility. Most of the portfolio effect appears with the first three securities.
  4. kevmay21



    So basically try to limit your risk as best you can so you can sleep with it at night.

    Try to diversify as much as your capital will allow.

    Try to add markets that are not very correlated.

    I will keep that in mind. I do not see a way of not going over the 2-3% risk area with a starting capital of around 40k though. It looks like its just going to happen.
  5. Divide the 40K into two parts of 20K each.

    Only trade reliable and repeatable stocks using leading indicators of price.

    At first, enter without rushing and then exit when a new trade presents itself even if you haven't peaked on price for what you own.

    Go to four streams of capital when you double ( about 8 4-day tradesof 10% eachin each stream) and have 80K rolling for you.

    If you think a stock is only worth putting 3% of capital into do not do the trade, ever.

    Only enter after you have had two leading signals of price occur and the price begins to BO as you go in.

    If the BO begins to fail you wash out with a small profit and go to the next opportunity.
  6. 2% risk model is based on trend-following diversified futures market, mainly from the Turtle era.

    Each strategy/system/model should be given it's own risk model.
  7. @kevmay21,

    more important than fixed "%"-rules is a feeling or prediction of your estimated profits and (more important) estimated drawdowns.

    I use monte carlo simulation software (an extended backtesting) to calculate the possible profits and account drawdowns of my systems. The max. account drawdown and the margin of the traded market makes ist possible to estimate the needed trading capital.

    But the so-called "system simulation" only shows the range of possible profits and drawdowns under the assumption, that your system results remain constant, what correlates in many cases with unchanged market conditions.

    An additional use of "data simulation / data scrambling"-generated synthetic test data and again new system tests allows also to test the systems under changed market conditions.

    These "simulation things" I do with a self-developed monte carlo simulation software. If you are interested see here:

  8. kevmay21


    Thanks all.

    Yes I have been doing monte carlo simulations to see pobabilities at diferent confidence levels. I use MSA Market Systems Analysis from Adaptrade.com.

    That program also lets me test a lot of position sizing schemes.

    I guess the whole thing that got me scratching my head is this whole 2-3% rule. Seems like it would be pretty hard to find a system with a largest loss of $800-1200

    I guess your average loss and up to 2 standerd deviations would be more important that that very rare larger loss.
  9. @kevmay21,

    better forget percentages and standard deviations.

    Take the absolute returns of the monte carlo simulation runs.

    In my software I analyse both the absolute and the "VaR" (value at risk) based results. And you will see, that VaR isn't a good choice to measure your trading system risk... :)

  10. kevmay21


    Clarification. I am getting 800-1200 based on 2 to 3% of 40k
    #10     Dec 14, 2006