Do Indicators lag?

Discussion in 'Trading' started by Daal, Oct 19, 2002.

  1. Daal

    Daal

    Aren't they ''leading'' indicators because of self-fulfilling profecy?
     
    #21     Oct 19, 2002
  2. If you can make money with them; who cares ...
     
    #22     Oct 19, 2002
  3. bone

    bone

    You know, Pabst, I can never find Pabst at the liquor stores these days. Don't make me drive up to Milwaukee.
     
    #23     Oct 19, 2002
  4. Pabst

    Pabst

    Bone,I bet there's alot of days when 94 is faster from the North Shore to Milwaukee than the drive to the Loop. :)
     
    #24     Oct 19, 2002
  5. bone

    bone

    Amen, brother.
     
    #25     Oct 19, 2002
  6. Good points,good as gold.

    There is a sense in which a good indicator like momentum lags-ever seen a bunch of momentum signals on your brokerage statement???????????????????????[mostly price on mine]

    --------------------------------
    Got to go,Ill try to dig around this gold mine later.:cool:
     
    #26     Oct 21, 2002
  7. All indicators lag because traders tend to oversmooth the lines in a desperate attempt to make them look pretty. Too many traders equate a smooth line for comforting certainty and a squiggly line for icky randomness. That is why everyone applies so much smoothing to everything. But in trying to plaster over the squiggles and shocks that prices are err to, they throw the baby out with the bath water (sorry for mixing mangled Shakespeare with a trite cliche).

    The entire notion of "smoothing" presumes that the underlying variable has some nice constant value that is hard to see because of random noise. Applying a moving average or smoothing filter helps smooth out the squiggles, but it works if the underlying true value is unchanging. If the true value of the underlying is varying continuously (i.e., the first derivative of the value with respect to time is defined and finite), then smoothing generates a lagged value. Using a zero-lag smoothing filter will help add the lag back in. But zero lag filters do presume that the value of the underlying is not varying too aggressively (best to look at this in frequency space and attempt to estimate the power spectrum of the true value of the variable).

    If the true value of the underlying is varying discontinuously (i.e., is subject to jumps in its value), then the smoothed value takes even longer catch-up to the true value. Worse, any attempt to use a zero-lag smoother will create nasty overshoot. A zero-lag filter will interpret a jump in the value (e.g., a large price gap) as evidence that the velocity or momentum of the signal is extremely high. A zero-lag smoother will over-compensate and push the estimated smoothed value far above the true value. Some adaptive smoothers explicitly detect jumps in the signal and handle them a bit better.

    But what if we embrace randomness by using totally unsmoothed (or minimally smoothed) indicators? This perspective makes sense once we see that smoothing adds error (in the form of lag) even as it removes error (in the form of random noise). Ideally, we would tune the smoother to balance the error added by lag and the error remaining in unattenuated noise -- finding some happy minimum of total error.

    So, what is good balance between the relative amounts of error added by lag and the error remaining in unattenuated noise? I would argue that error from lag is far more dangerous than error from noise when it comes to trading. Error from noise would slightly randomize entries and exits -- sometimes the unsmoothed indicator would get the trader into a position a little too early, sometimes it would enter a little too late. In contrast, error from lag is consistent and would lead to consistently early or late entries and exits. A random influence on the P&L of each trade is not so bad because it averages out over many trades. A consistent error in the P&L of each trade is more likely to contribute to nasty and prolonged drawdowns.

    The point is that if you embrace randomness, you can create trading systems that work just fine on un-smoothed or minimally smoothed indicators. Pretty smoothed lines do not equal certain profits.

    Cheers,
    Traden4Alpha
     
    #27     Oct 21, 2002
  8. dbphoenix

    dbphoenix

    I'm quoting this because it is such an important point and it should not be missed.

    --Db
     
    #28     Oct 21, 2002
  9. Momentum is rightly nicknamed by Goldenarm a leading indicator. Unfortunately it's also a leading indicator when it comes to premature ,goofy exits on a good trend.:) [smiley a traden4Alpha quote.]

    Have seen some market conditions and sectors and time frames where it's helpful.[momentum].

    ----------------------------------
    On a swing trade or profit max daytrade, momentum in my opinion , can be more useless dirt than gold.

    Momentum=price speedometer. Many moves [including non trending] slow down ,but go on and on and on.:cool:
     
    #29     Oct 21, 2002