BACKTESTING: only 25% winners on MA crossovers?

Discussion in 'Automated Trading' started by clambill, Mar 8, 2009.

  1. Uh, is it normal to only see 25% winners max on moving average crossovers? This is the first time I tried backtesting. And I did it on DealBook 360 (from GFT Forex). I find that a little low.
  2. auspiv


    ma crossovers are not a winning strategy
  3. look at it the good way, if you fade (do the opposite) the MA crossovers, you get 75% winners.

    isnt trading awesome?
  4. With the same exit parameters + slippage, not quite. Both can be losers (and usually are). And by the way, there are two sets of results to be concerned with: win rate, and win/loss profit ratio... 25% can yield you success if you win 5x more than you lose per trade on avg.
  5. FaceOff


    I did a substancial amount of testing on multiple MA's on 2+ years of data over a range of contracts to see if there was any substance to the claims. Tried a bunch of different combinations and permutations with most suffering when quick reversals occurred. I did however find that a moderate amount of profit (only 350 points over 2 years) could be squeezed out of the ES with a 1hr bar. So there are isolated classes where you can get it to work.

    Also, the exit strategy will significantly impact results. Try using a trailing stop, rather than exiting when there is an entry signaled in the opposite direction.
  6. What is better?
  7. auspiv


    that's for you to figure out
  8. No, you need to win slightly over 3x more than you lose per trade on avg. (these avgs include of course all costs in actual trading).

    The equation is: W = PF/(PF+R)

    where W is the win rate, PF is the profit factor and R the avg. win to avg. loss ratio.

    Solve this for R and you get

    R = PF x (1-W)/W

    Plug in PF =1, W = 0.25 and you get R = 3

    so if R > 3 you are making money. If R = 5 and W = 0.25 then your profit factor is

    PF = R x W/(1-W) = 5 x 0.25/0.75 = 1.67 and you are making a lot of money:)

    A derivation of this formula can be found here.
  9. MA Crossovers are bunk.

    If you don't believe me, bet BIG on them. I could use the cash.

    Or you could waste $5,000 - $10,000 on Vantagepoint "Intermarket" Analysis Software, out of Wesley Chapel, FL.

    :D :D :D :D :D
  10. I've done a lot of work. If someone helpful would say something like: "fib ratios" or ANYTHING helpful, I'd be willing to consider ideas. But, that kind of reply, I don't need.
    #10     Mar 9, 2009