Probability of success due to chance alone?

Discussion in 'Risk Management' started by anesthesiaman, Oct 4, 2010.

  1. Random entry and exit, yes but should the iterations trades not have similar trade duration as your real trades ?
     
    #21     Oct 5, 2010
  2. You've done good work. But think about it. If you risk 38% of your capital of 100K and you lose once you are left with 62K. If you lose a second one (this is approximation because w and R have changed) you are left with 38.44K. If you lose a third one, you are left with about 23K. Would you like that? Three consecutive losers are very common in trading. My recommendation: stay away from %Kelly. Also, this is a good article on %Kelly by the same author as the other one.

    As far as your numbers, your sample is very small to reach any conclusions, although you have done very well. As far as running simulations to determine, or better said to test the hypothesis that your results were achieved by chance alone, keep in mind that you are not dealing here with stationary random variables having known distributions. You are dealing with non-stationary random variables with unknown distributions. Even if you determine via simulations that these results have 0.01 probability to be the result of chance, that does not mean anything. Neither 0.001 means anything. This cannot tell you whether you obtained the results using skill or chance alone. You need another measure which will never be available to you. The relationship between p-values and skiill. Nobody knows this relationsship but time and number of samples allows us to say something about it. For example if you have 1,000 trades and consistent returns with small drawdown for 5 years, then we can say with high probability - except in a set of measure zero - that your results are due to skill/edge. In other words, even in that case there is a remote probability of chance but we cannot know the value of that probability.

    I personally feel that you have some skill.
     
    #22     Oct 5, 2010

  3. Can you elaborate on this ?
     
    #23     Oct 5, 2010
  4. As Bill said, your sample is much too small.

    Don't base your performance on the performance of others, especially individuals that have an effect on liquidity in the marketplace, they are not your "mark".

    I didn't read everything else in this thread, but I do have some advice. Forget about all these statistics and p values etc. This random probability bullshit and all of that.

    If you are not running a specific strategy there is no possible way you will ever come upon what you are looking for, this holds especially true if you are 100% discretionary.

    I recommend keeping up with your medical career instead as I can give you a 99.99995% probability that you will be more successful in that rather than trading futures.

    Gl though.
     
    #24     Oct 5, 2010
  5. IMO, financial data are best described by a random walk with drift plus a stochastic trend like in

    X (t) = a + X (t-1) +b(t) + n(t)

    where a is the drift, b(t) is the stochastic trend and n(t) a white noise term

    This is a non-stationary process, i.e. mean and variance vary with time.

    Although there are conflicting studies, some works indicate that in the case of S&P 500 Index, the main cause of non-stationary returns is the time-varying second unconditional moment.

    In simple words, markets change behavior over time. Find what, if anything remains constant, and profit.
     
    #25     Oct 5, 2010
  6. HA! I KNEW there were some smart folk on this message board that know their statistics pretty darn good.

    And alright.... as other people have shown me in mumbo jumbo statistics terms, perhaps I was searching for the wrong thing, a p-value, which doesn't really help my trading anyway. I was simply looking for it because I wanted to know if I was just extremely lucky or had at least some trading skill.

    Having been focused on medicine all my life and not having a financial background means I just don't know what to or how to compare (or what to expect) from trading other than my P/L at the end of the month.

    Although now after having a good discussion with you gentlemen/gentleladies, at least I have been exposed to certain new ideas (to me) and also these formulas. For example, perhaps I should calculate my, uh, expectancy - even if the sample size isn't over several years (attempting to sound financial and non-medical).

    Even if I haven't been trading since 2003, when I finally learned how to do it profitably but was focused on medicine, I have been watching closely the market - due to the love of it and all the new data that always tweaks the outlook.

    It is a constantly changing story, filtered through my own personal market view - but with familiar, repeatable patterns - that make it a fun to trade as a semi-professional trader!

    There is much to learn and I will look into all these books and formulas that have been mentioned so far simply - just for the pure joy of learning it - perhaps that is why I've been lucky so far? I know I don't like to watch sports and can't keep a decent conversation in NFL, NBA, MLB or any TV but devour WSJ and Bloomberg on my iPhone. There is too much to learn (personal trading learning and market outlook learning i.e. news) and too little time. Sports watching just prevents me from learning more about the markets.

    Ironically however - the simplest approach seems to create my success. NONE of these values or calculations of expectancy, R-value multiples, nor Kelly % tell me what I don't already know: maximize my profits and limit my losses (in my case this means stop selling so soon and minimize my hedging - all within the context of recognizing my own emotions/anxieties during the trade).

    I would argue that I do have at least some skill in discretionary trading - I just don't know how much.

    With respect to my trading colleagues,
    Anesthesiaman

    Recent results (disregard DODGX it was one share I bought years ago - when they reopened the mutual fund and I wasn't sure if I wanted in but didn't want to miss out so I just bought one share - and sold it this year). Some of my losses are hedges and designed to lose - others like wheat - are too early of an entry point.
     
    #26     Oct 5, 2010
  7. I think some of the intervening posts have (re)stated it well. You are basically looking for value where there is too little, too many details left out, and too little grasp of the statistical and market realities.
     
    #27     Oct 6, 2010
  8. olias

    olias

    I think this is an interesting statement. I don't think than any particular period can be labeled as 'easy' or 'difficult'. The market is always the same: very difficult
     
    #28     Oct 6, 2010
  9. A rising tide lifts all ships...if the broader markets are all going up in unison, it is what I would consider easy. Remember 1999?

    When the VIX is high however, as this past summer was, does it become more important (i.e. more difficult) to get entry and exit points more timely.

    High VIX is like anesthetizing a sick pt - it requires much more vigilance and careful titration of induction agents.

    But according to some, I am just unskilled opinion. So in fact disregard everything I've said. Because anyway - the yin-yang of it is: I know nothing.
     
    #29     Oct 6, 2010
  10. How does "risking" 38% of capital mean he will lose the whole 38% if the trade is a loss?

    I may be mistaken but doesn't Kelly look at how much you should "bet" and not how much you should risk?

    As in he could allocate 38000 to a trade but then as part of his set up he could have a 10-20% stop loss or a stop of some other sort. Or something like that.

    Can somebody please clarify?
     
    #30     Oct 6, 2010