Risk Of Ruin Calculation

Discussion in 'Technical Analysis' started by jtTrader, Feb 21, 2007.

  1. jtTrader


    This is the results from excel where I coded the formula from page 68 of Kaufman's "Smarter Trading". It does not seem possible that this system has a RoR of 21%.

    Risk Of Ruin

    AvgWin 1145
    AvgLoss 355
    Investment 200000
    % Profitable 0.6429
    % Losing 0.3571
    MaxRisk 0.0025
    AvgWin % 0.0057
    AvgLoss % 0.0018
    Z 0.0030
    A 0.0037
    P 0.9079
    Risk of ruin 0.2162

    I am assuming I am not handling the captial units correctly. Could someone check these numbers or point out my error? I have double checked the formulas, and I can't find the mistake, but these numbers seem wrong.

    (I can add the excel if necessary)

    Thanks for the help
  2. jtTrader


    I attached the spread sheet with the RoR formulas from smarter trading. Anyone see the problem?

    Thanks again,
    • ror.xls
      File size:
      14.5 KB
  3. kut2k2


    RoR only applies when you use an absolute position-sizing method, not a less-than-100% fractional position-sizing method. Use the latter and your RoR worries are over. :cool:
  4. jtTrader


    All the RoR samples I have seen are always futures oriented absolute position sizing. I can't recall seeing a stock trading based example.

    I trade 2000 share lots as 1 position, but the share price can vary from $15 - $100, but position risk is the same.

    How would I account for that, to get an idea of RoR on my systems?

    Thanks again,
  5. PetaDollar

    PetaDollar Moderator

    Hello ET'ers, let's feature this thread for a while. It is not TA per se, but is intimately related in my opinion, because the numbers that go into finding RoR come from TA (my definition of TA = analyzing historical data for the purpose of estimating probabilities and other quantities important to successful trading).

    I would like to point out to those marching towards (but have not yet achieved) profitability that devoting a great deal of attention to something like RoR is a key ingredient. Calculation of RoR by the book is one way to go; but the main point is to clearly define a line in the sand between bad luck and a strategy that doesn't work. Futhermore this is not something for just system traders. Even someone who trades without a plan at all (not recommended) can look at past trades and make some probability estimates. The transformation to profitability requires an equal amount of attention to all the important aspects of trading (for example, RoR), not just entries (what most newcomers fixate on).
  6. I personally like the idea of using ROR for risk measurement (see my thread and latest post: http://www.elitetrader.com/vb/showthread.php?s=&postid=1372197#post1372197)

    The problem I'm running into though is I don't have any true measure to compare to so I don't know if it's accurate or not. It makes sense to me to use it though since it only requires a single pass and isn't based on random trials, but I thnk I'm at the point that I need the random trials to at least validate the numbers.

    I'm curious though how many people use ROR in their risk analysis.

    For the original poster if the strategy just uses single trades I'd use the actual trade data - in particular the excursion data for each trade to get the max swings - and determine the ROR with the formula I linked to earlier.
  7. Yes, your formula for A in cell B12 has been entered incorrectly. Check parentheses against Kaufman. Your A should be 0.0047, P=0.8233, RoR=.4419=44.19%. (Makes even less sense, I know.)

    Second, Example 3 on p. 69 of Kaufman has errors. A should be 0.0205, P=0.6464, RoR=0.0006=0.06%. Not 44.1%... minor details, right?

    Third, Kaufman's formula itself cannot be correct as printed: the higher MaxRisk, the lower Risk_of_Ruin. That's the opposite of what you'd logically expect, for any positive-expectancy system. Not a chance.

    I own neither Ralph Vince's book "Portfolio Management Formulas" (1990), from which Kaufman states he got those formulas, nor Vince's source, P. Griffin, "The Theory of Blackjack" (1981), in order to verify. According to another (excellent) book on my bookshelf, Nauzer Balsara, "Money Management Strategies for Futures Traders" (1992), p. 16, there is no exact, closed-form solution for risk of ruin, when the average win size is not equal to the average loss size. In the real world, that would be... (drum roll)... 100% of the time. Monte Carlo simulation is the solution offered in that chapter.

    Is Prof. Balsara wrong? I don't know. MC in Excel is what I've always used to evaluate various MM algorithms, estimate drawdowns, etc., as it's infinitely more flexible and powerful than any formula could be.
  8. Fractal


    Excellent post.
  9. promagma


    Your ROR will be very low .... but more important is how many trades are in the sample size, and does it maintain performance with out-of-sample data and different market conditions. EricP has some nice formulas -


    Stats on 50 trades means nothing. If that is 500 trades you have a killer system :D
    #10     Feb 28, 2007