Which is the most important stat to you?

Discussion in 'Strategy Development' started by TradeNYSE, Aug 1, 2007.

  1. Working with my current trading strategy and trying to optimize the system. Assuming there is one variable that can directly affect the number of trades, profit/loss, which is most important to you and why?

    For ex. (just to clear up)
    Let's say if 1x was used as the variable the results were (YTD):

    Average profit/trade = .34%
    Median profit/trade = .30%
    Wins = 1731
    Losses = 1304
    Win % = 57%
    Avg Win = 1.79%
    Avg Loss = -1.60%
    Avg Win/Avg Loss = 1.07
    Total Buying Power Required for all trades for entire year = $60,740,000
    Profit = $102,611

    Now the variable is 2x:
    Average profit/trade = .51% (+.17% higher than 1x)
    Median profit/trade = .49% (+.19%)
    Wins = 1061 (less trades triggered)
    Losses = 719
    Win % = 60% (+3%)
    Avg Win = 1.99% (+.20%)
    Avg Loss = -1.68% (-.08%)
    Avg Win/Avg Loss = 1.19 (+.12)
    Total Buying Power Required for all trades for entire year = $35,600,000
    Profit = $90,529 (roughly -$10k)

    My question becomes...are you more concerned with the bottom line (where 1x profits more than 2x), or are you more concerned with another ratio/number? What ratio/number/stat would you focus on (I do not have a statistics background, so I am sure I probably left out some very crucial calculations that could help me...that is where I need some input).

    I mean from a common sense standpoint, I can look at this and say that 2x is a much better/efficient way b/c you lose $10k in profits, but you risk almost half the money in buying power over the year. But What if I have 50 different variables and want to pick the best one?


    Thanks in advance.
     
  2. Brandonf

    Brandonf ET Sponsor

    Hips, waists, and breast?

    On a more serious note though, I tend to be more concerned with my drawdowns. How large are they vs my upswings. How long does it take to recover from a drawdown? How much of my return depends upon the market? Things like that are important to me. Plus Hips, Waists and Breasts.
     
  3. Thanks Brandon...but looking for something more statistical and not discretionary.
     
  4. Roscoe

    Roscoe

    I use Vince's Weighted Geometric Mean as my primary stat along with SDevRatio and a few distribution stats.
     
  5. Have a good idea of how and why the largest losses occur. This will reveal the biggest flaws of your method and keep you from being on an unnecessary roller coaster ride.
     
  6. Ok let's try this again:

    System 1
    TOTAL PROFIT $154,626
    AVERAGE/TRADE $57
    MEDIAN/TRADE $58
    WIN 1632
    LOSE 1068
    WIN % 60%
    AVG WIN $220
    AVG LOSS -$192
    WIN/LOSS[abs value] 1.15
    TOTAL BUYING POWER REQD $55,000,000
    RETURN (profit/capital used) 0.29%
    StdDev of the trades $294

    System 2
    TOTAL PROFIT $166,573
    AVERAGE/TRADE $46
    MEDIAN/TRADE $49
    WIN 2130
    LOSE 1484
    WIN % 59%
    AVG WIN $206
    AVG LOSS -$184
    WIN/LOSS[abs value] 1.12
    TOTAL BUYING POWER REQD $73,000,000
    RETURN (profit/capital used) 0.23%
    StdDev of the trades $278


    Which would you choose and why? Are there other calculations you would want to compare that are not listed that would help you decide?
     
  7. Didn't answer my question, but thanks for the input.
     
  8. None of the above. I like to look at sharpe/sortino ratios. This gives you a good idea of the risk/reward.
     
  9. Can someone help me with what I need to do with my data to get the numbers required to calculate the sharpe/sortino?
     
  10. Sharpe = (Annualized Rate of Return - Risk Free Rate of Return)/Annualized Standard Deviation of returns.

    Sortino = (Annualized Rate of Return - Risk Free Rate of Return)/Annualized Down Standard Deviation of returns.

    I don't know what you need to do to your data to calculate these, since I don't know what your data contains.
     
    #10     Aug 2, 2007