Distribution of trading profits: What is it?

Discussion in 'Trading' started by mdlm, Apr 20, 2002.

  1. mdlm

    mdlm

    What does the per trade dollar P/L distribution look like? Gaussian distribution? Uniform? Binomial?

    My guess is that a successfull trader's distribution looks like a bell curve with the left tail truncated (because losing trades are cut short) and a fat right tail (because every once in a while the trader has a very profitable trade).

    Does anyone have evidence from their own trading? If you have the per trade dollar profit stored in a file, you can plot a histogram in Excel using the histogram function (Tools/Data Analysis/Histogram).

    I am particularly interested in hearing from traders who have access to real trading data.

    (Note that if you trade different size depending on the trading signal, the stock, volatility, etc. you should not correct for this. However, if there has been a broad offset in your trading, from say 100-500 shares per trade to 1000-5000 shares per trade, then you should normalize for this.)
     
  2. JUST TRADE AND MAKE MONEY!!
     
  3. tntneo

    tntneo Moderator

    LOL :D
    It's important to focus on the real thing. becoming constant winner is already tough enough, who cares about equity curve analysis ?

    seriously, you bother about that only when consistent for a while. it is important to analyse your own trading in order to improve it and monitor your business. that helps detecting when conditions are changing or something is wrong. when everything is OK well just too bad.

    to reply to the original question, yes it is bell curved. And it makes sense. to be consistent in making money, you are following a method (or a few). a method will have a typical reward, typical loss etc.. so the gains are distributed normally.
    investors may have different distribution but they don't usually have a method anyway. I am replying regarding traders.

    tntneo
     
  4. sabena

    sabena

    The answer to your question , you can find

    on www.snailtrader.com under the topic

    "snailtrading:playing probability"
     
  5. sabena

    sabena

    The better your system, the smaller the ratio

    standard deviation / average trade.
     
  6. sabena

    sabena

    "The true nature of things we will never know,
    NEVER" - ALbert Einstein.
     
  7. nitro

    nitro

    Sabena,

    I enjoyed this (your ?) site.

    nitro
     
  8. MDLM,

    The optimal shape will be a normal distribution with a fat right tail and a truncated left tail ...

    Dissection:
    a) a variant of any normal distribution implies the majority of trades will be small winners and small losers
    b) a fat right tail is indicative of running profits
    c) a truncated (not thin, but truncated) left tail is indicative of sound risk management
     
  9. Well that's just fantastic advice! NOT. That's about as helpful as telling as sick man to GET BETTTER or a shy guy to BE CONFIDENT.
    The thing is, HOW? What steps can we take to get better, be confident, or MAKE MONEY?

    Will studying the distribution of your trades help you trade better? I say it can. Besides, mdlm never said, "help, I'm losing money", did he? Maybe he just wants to make more.

    Tnteo, I don't think mdlm is talking about "equity curves". And just because it "makes sense" to you that "gains" (I assume you meant "trades") are normally distributed, it doesn't mean that that is actually the case. (It once "made sense" that the world was round.)

    What if I entered and exited the market at random? I bet you I could come up with a heap of distributions that ARE NOT normally distributed.

    I believe that studying the distribution of trading results is most useful when backtesting a mechanical system, because it allows us to draw inferences from the data and make adjustments to the sytsem. (Which is NOT the same as curve fitting the system.)

    If you do this sort of thing you'll find that most (99.9%) of systems will not be perfectly normally distributed, although many will be close.

    Desirable characteristics of a distribution would be a positive skew (ie, skewed to the right) and a negative kurtosis (meaning a fat body with thin tails - IMHO anyway, I'd prefer a system that doesn't stand or fall on outlier trades). For a more thorough discussion on this topic (and trading systems in general) I would refer you to the work of Thomas Stridsman.

    I don' t know how to post the data from trading results here (and I'm not sure that I'd want to either!), but PM me and I'll email it to you.

    Daniel
     
  10. Daniel M,

    Point taken... hopefully the Elitetrader Trading Style Poll on another thread will, in due course, benefit from comments from members on the relative merits and demerits of the plethora of potential trading styles out there...

    Regards,
    Candle
     
    #10     Apr 20, 2002