More Winners or Larger Avg. Winner?

Discussion in 'Strategy Development' started by Corso482, Jan 27, 2003.

  1. Assuming the two result in the same return, would more winners be superior because it result in a smoother equity curve?
  2. AusTrader


    Theoretically more winners will smooth the equity curve. In real life however a higher win rate tends to cause a lower avg win. Larry Williams is a good example. Win/loss ratio is low so you can do lots of good trading and then get thumped by one nasty loss. Better to go for 50% win rate with a higher avge win/loss, say > 1.5
  3. A system with more winners (on an intraday basis) usually equates with scalping.... the equity curve can be smoother, but it is absolutely vital to factor in the impact of enhanced aggregate commissions on scalping strategy expectancy... the way to offset the commission effect to equalize scalping expectancy with expectancy on a strategy with larger winners (and larger drawdowns) is to necessitate a very high hit rate on the scalping strategy...

    Theoretically (via backtesting), the acquisition of a very high mitigating hit rate on a scalping strategy is often feasible, but you must be sure to make sure that there are no practical obstacles (execution issues) in the way of the manifestation of the theoretical expectancy...
  4. candletrader,

    sorry. i was talking about swing trading.
  5. My apologies... well, in the case of swingtrading, the commission effect is negligible, so you would not need an inordinately high hit rate on your "more frequent winner" strategy... personally, I could not (from a psychological perspective) absorb the larger drawdowns of a "larger winner" strategy, so I would go for more winners...
  6. acrary


    If both strategies have the same expectancy, the same number of trades, and you're using a % of equity for your risk sizing method, then the higher win % is superior. It will have a lower DD, higher return and a better risk:reward ratio.

    Here's two identical strategies:

    Method A

    % win = 40%
    win size = 3.375

    % loss = 60%
    loss size = -1

    expectancy = (.4 * 3.375) - (.6 * -1)
    = .75

    Method B

    % win = 70%
    win size = 1.5

    % loss = 30%
    loss size = -1

    expectancy = (.7 * 1.5) - (.3* -1)
    = .75

    I setup 100 trades of each method in their win/loss proportions. Then I ran a Monte Carlo test with 1000 randomized passes across the data. Starting equity was $100,000, risk was 1% per-trade, and the multiplier for the pts. was $50.

    Method A

    Max. DD ($14,850) (1 in 1,000 passes)
    DD at 95% level ($6,381)
    DD at 50% level ($1,000)

    Lowest ending profit +$104,843 (1 in 1,000 passes)
    lowest 5% of profit +105,581
    50% level of profit +106,431

    Reward (at 50% level) to Risk (DD at 95% level) = 106431/6381 or 16.68

    Method B

    Max. DD ($4,950)
    DD at 95% level ($2,000)
    DD at 50% level ($0)

    Lowest ending profit +$108,950
    lowest 5% of profit +109,300
    50% level of profit +109,750

    Reward:Risk = 109750 / 2000 or 54.88

    The lower DD's are caused by the higher win %. The higher profits are caused by more of the compounding opportunities taking place while making new highs instead of digging out from the DD.
  7. I have only ever read two of your posts acrary, and both have been phenomenal... thank you :)

    P.S. your ptsp gif file was fantastic...
  8. The original question says both methods have the same return. As such, I don't think your analysis applies to this specific question.

    The question seems to be centered on whether a smooth equity curve is better than one less smooth.

    To me, that is merely a matter of preference, ie., suitability.

    Again, this is based on my presumption that return is equal between the two methods, as stated in the initial post. If return is equal, I am presuming further we are talking about net profit after commission and slippage, say at the end of 12 months. If this presumption is true, then drawdown is irrelevant.
  9. inandlong,

    Is the above some sort of 'absolute' truth you are stating or just your opinion? :confused:

    Perhaps to YOU DD's are irrelevant, but certainly not to ME...

    Simply looking at the end result without taking into account the path it took to get there is no better than being blinded by the linear micro-level path it takes to get to the macro-level result...

    I personally never look at just the return -- it is always DD/NET ratio...
  10. I agree, have parameters like that in two of my systems, the third one is different, but it still has a very nice equity curve, or rather has had for over 12 months with a very shallow dip only in one of those months. But in 2001 the equity curve was not so pretty.
    #10     Jan 27, 2003