Risk Of Ruin Calculation

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

  1. hcour

    hcour Guest

    I posted threads in 2 different forums asking what statistics traders thought were helpful but didn't get much feedback. I'll give it a shot here. What others, besides ROR, and the basics such as WIN%, AVG WIN, AVG LOSS, would be good for a discretionary trader to monitor?

    Thanks,
    Harold
     
    #11     Mar 1, 2007
  2. Ruin is a 100% drawdown.

    If you get wiped out... you are likely done for good...
    So I don't see how ROR calculations are relevant...
    Because your ROR should be so small (way < 1%) as to be irrelevant.

    My biggest drawdown in 13 years is 10%...
    And I was leveraged 6:1 for long periods...
    So my Risk of Ruin is, for practical purposes, zero.
    The ROR for any well managed firm... is zero.

    ** Minimum standards ** for success as an independent trader:

    Consistent annual return in the 20-30% range...
    With maximum expected drawdowns of under 20%...
    Which would result in a Sharpe Ratio in the 2.0 to 3.0 range.
    My 2 best years I had a Sharpe Ratio > 4.0... so it can be done.

    You can argue with this...
    But you cannot build a business on volatile returns.
    If one thing doesn't get you... another will.

    Become a compulsive, obsessive hedger.
    Never make directional bets... that is for gamblers.

    And 50 or 500 trades is a joke.
    Many small firms do > 500 trades/day.

    I find that the poker players at 2+2 Forums...
    Have a ** far better ** understanding of variance Zero Sum Gamers encounter...
    Than the traders at ET.
     
    #12     Mar 3, 2007
  3. kut2k2

    kut2k2

    Utterly meaningless. Poker has exactly calculable odds. Trading doesn't. That's why most traders have no idea how to apply optimal position sizing (which is also zero RoR, btw) to trading.

    Ask your **far "smarter"** poker players to do it for trading, and see how fast they fold. :D :D
     
    #13     Mar 3, 2007
  4. The problem with your response...
    Is that you ** completely ** miss the point of my post :D :D

    VARIANCE, VARIANCE, VARIANCE...
    Which has zero connection to whether poker has "calculable odds"...
    As opposed to another Zero Sum Game.

    To give a specific example...
    A common way to make 50-100K/year playing online poker...
    (Was 2 years ago... and perhaps still is)...
    Was to multi-table SNGs using concepts from the "Independent Chip Model"...
    And play about 30-50 of these 45 minute one table tournaments every day.

    Now many NEW players would play 100 tourneys = roughly 100 hours of play...
    And be making a 20% return on investment...
    And be all thrilled and swagerring about their "success".

    But ALL the winnings players at 2+2 Forums would just laugh...
    Because they knew 100 hours of play is ** totally meaningless **...
    Because variance/luck completely swamps your results.

    In fact, using simple high school math and VB6...
    It took me about ** 30 minutes ** a couple of years ago...
    To calculate confidence intervals for SNG play...
    If you are doing a 20% return.

    Here is an Excel workbook that shows the results... not the calculations:

    http://www.pathcom.com/~gzt/SNGVariance.xls

    The table shows the following:

    If you play 100 SNGs = 100 hours of play and do 20% return...
    You 95% confidence interval is 20% profit +/- 36%.

    If you play 1000 SNGs = 1000 hours of poker play and do 20% return...
    Your 95% confidence interval is 20% profit +/- 11%.

    Bottom line...
    It takes at least 1000 hours of online poker to even establish that you are a "winning player"...
    And even after 1000 hours your profit level can only be approximated +/- 10%.

    The winning players at 2+2 Forums ALL understand this intimately...
    And this is hardly rocket science...
    It's f*cking high school math.

    Stock traders are subject to a roughly comparable variance as poker players.

    What is the trading equivalent of 1000 hours of poker play?
    Definitely not 50 or 500 trades.
    Depends on your trading/hedging style.

    In fact...
    Only traders are subject to potentially catastrophic variance like we saw last week.
    Nothing comparable to a market crash or Enron bankruptcy exists in poker.

    If you cannot do this kind of variance analysis...
    In one hour pretty much off the top of your head...
    Then you are completely out of your depth in addressing the mathematics of trading.
     
    #14     Mar 4, 2007
  5. tireg

    tireg

    Just a correction to the calculation.

    The interpretation for '95% Confidence Interval' is for the variation of the mean return (here, 20%) around some true mean return.

    The confidence interval here is actually much smaller than your initial estimates.

    Further explanations:
    http://www.andreassteiner.net/perfo...n_Calculation:Advanced_Topics:Average_Returns
    http://en.wikipedia.org/wiki/Confidence_interval
    http://davidmlane.com/hyperstat/A103735.html

     
    #15     Mar 4, 2007
  6. You know...
    I don't have a lot of time for a debate on this.

    The reason I believe my post to be correct...
    Is that it ** confirms ** what the smartest guys are posting on 2+2 Forums...
    Which is a far, far smarter place than ET.

    ** Intuitively **...
    My numbers appear too high...
    (That's why I did this analysis 2 years ago)...
    And your numbers appear to be way too low...
    But there is an exactly correct answer.

    You have to run it on all the possible returns for a single SNG tournament.

    For an $11 Party Poker SNG... 10 people pay $11:

    1st place gets $50...
    2nd place gets $30...
    3rd place gets $20...
    4th to 10th place gets MINUS $11
    And Party Poker keeps $10.

    So if you are average player...
    You can program a long series of data points that go 50, 30, 20, -11, -11, -11, -11, -11, -11, -11...
    And extend/repeat that, say, times 100 for 1000 games...
    And you will see what the 95% Confidence Interval for that data is.

    One can similarly simulate data for a top player that usually wins more than that...
    Typically a Top Player has a 20% profit margin.

    The confidence intervals...
    Are definitely higher than one would expect intuitively.

    For me I don't care...
    Because I make about 75,000 trades/year...
    And variance is not an issue for me.
    (It would be in poker).

    But for people starting out with $10,000 or $20,000...
    And doing high risk trading with a Risk of Ruin of 20-30%...
    (Been there... done that 20 years ago)...
    It's extremely important to understand the variance of the game they are playing.
     
    #16     Mar 4, 2007
  7. ha-ha-ha...
    No one noticed...

    But the data points for an "average player" are not

    +50, +30, +20, -11, -11, -11, -11, -11, -11 = +23

    But actually

    +39, +19, +9, -11, -11, -11, -11, -11, -11 = -10

    If an "average player" plays 10 $11 SNGs...
    He breaks even against the other players...
    But Party Poker relieves him of $10 in rake...
    Which is a profit margin of -10/110 = -9.09%
     
    #17     Mar 5, 2007