If there is a quantitative measure for Edge, what is it?

Discussion in 'Strategy Development' started by OddTrader, Jun 9, 2006.

  1. Can we measure Edge?

    Do you have anything in mind that could be used to measure an Edge with quantitative terms?

    Any pointers?

    All comments would be appreciated.
     
  2. Page 135 - "An edge is nothing more than an indication of a higher probability of one thing happening over another." (thanks wareco)

    If you agree with the above statement, then I would say over a series of trades you would have a positive expectancy.

    How many trades would it take for someone to feel confident in there edge, it would be different for every trader.

    Having said that, I lean more towards the ready, fire, aim, school of thought. I need to immerse myself in whatever I am doing. Its to easy to get all intellectual about something and then base your assumption on a thought and keep going down this road of assumptions and thoughts about how you would like it to be.

    In other words you can estimate that it will only take 3 hours to dig that ditch but until you get down to the actual process of doing it and running into roots or a waterline or whatever is beneath the surface it's all just an interesting mental exercise.

    Its kind of like swimming you could read a 100 books on the subject but until you got in the water and started flailing about then and only then can you start to apply some of the techniques you learned in the books.

    Now if you are a super trader and have already mastered yourself and can control your emotions and execute your edge as written then you already have your answer.

    Its similiar to starting a new business you have an idea, you don't know if it is going to be profitable, you have to try it out, take it for a drive, see how it feels to you. It has to align with how you see the markets.

    I went back and read your question again and I think I kind of veered off the road a bit. That will teach not to reply to a thread while I am sober.
     
  3. Are you saying something like this:

    EdgeExpectancy=WinProbxAveWinAmount - (1-WinProb)xAveLossAmount

    WinProb=WinTrades/TotalTrades

    TotalTrades>N

    N=Any number greater than say 1,000 (Whether that should be simulated or actual?).
     
  4. You make more money than you lose, that is the measure of an edge :D, why complicate things people?
     
  5. So with an initial capital (say) $1,000, you do believe making $1 after making 1,000 trades is an edge.

    Do you have the required confidence good enough to increase trading capital by adding (say) 20 times (i.e. $20,000) of initial capital in order to make more profits with this proven edge, considering the potential risk to be faced immediately?
     
  6. mhashe

    mhashe



    P is directly proportional to E
    L is inversely proportional to E

    where

    P = Accumulatied Profits over time period X
    L = Accumulated Loss over time period X
    E = Edge
     
  7. I believe that you are asking the following: "Are my results (backtested or otherwise) statistically significant?"

    i.e. you are trying to ascertain that your out-of-sample good (I hope - and by whatever measure you define good) results have a high probability of not being attributed to chance alone.

    Do a search on the web about statistical inference, hypothesis testing etc. Depends on your current stats knowledge.

    If memory serves, Aaron Schindler did a web presentation (CME, or CBOT?) a while back where he presents a trivial system and touches on the subject - that might be of some help if this subject is new to you. It is rather simplistic though - but a start nonetheless.

    [Edit] Found it:
    http://www.cbot.com/cbot/pub/cont_detail/0,3206,929+15785,00.html
    Can't remember how relevant it is...

    Then again, there is no guarantee. Spurious relationships exist - it can be extremely difficult to distinguish causal relationships (particularly ephemeral ones) from spurious ones.
     
  8. How about MaxDrawdown>90%, still calling an edge and keep trading the system with additional capital by 20 times? :confused:
     
  9. Hey, that $1 dollar can buy you a collect call, 4 bubble gum balls, or a 99 cent burger at MickyD's.

    I like it!!!

    Plus, if you make a thousand trades on 1,000 capital contribution, you deserve to at least make a dollar in my opinion. :p

    It's better to have won than to have lost, than to have to beg for your hamburger on the sidewalk :D.

    We gotta do.... what we gotta do.
     
    #10     Jun 9, 2006