this is my contribution to ET

Discussion in 'Technical Analysis' started by dafong, Sep 20, 2008.

  1. dafong


    here's what i'm working on, but i need your guys' help to improve it.

    the technique: every time stochastic %D is overbought, you mark the highest point for while it was overbought. every time stochastic %K is oversold, you mark the lowest point for while it was oversold. Still don't get it, look at the picture.

    Now, we know stocks move in waves, and trends form; uptrend = higher lows w/ the occasional higher highs and downtrend = lower highs w/ the occasional lower lows

    ok, now the strategy is that every time stochastic %D crosses above 30, mark the low for the oversold period that just occurred. Compare that low to the low that occurred the LAST time it was oversold. Is this most recent low HIGHER than the previous one? if Yes, go long.

    now, when stochastic %D crosses below 70, mark the high for the overbought period that just occurred. Compare this recent high with the higher the LAST time it was OVERBOUGHT. Is this recent high lower than the previous one, if yes, go short.

    i would like to hear what u guys think. the green lines i marked was when there were either lower highs or higher lows.
  2. dafong


    here's pic
  3. dafong


    ok, again...
  4. It's called divergence, and it can work.
  5. I wonder if there is a mathematical way to write these rules. Stochastic might not be necessary, a moving average might also work, or just the price difference from say 10 days earlier or maybe just the lowest low or highest high price value in a minimum 5 % price fluctuation.

    I might use a stop loss rule. What happens if price values decrease a lot without rallying?

    I think it needs a position sizing algorithm.
  6. moarla


    DIVERGENCE WORK SOMETIMES, sometimes not. In your example works nice, but i can post as many other charts where divergence fails....

    I have a system on Stochs, that is unfailable :)
  7. man


    so this is your "contribution" to et ... asking people if
    they can evaluate an idea of yours. i'd call that different.
  8. bbqbbq


    problem with your idea, it sounds nice, but you need a very refined entry system. with stochastics you down know if they are gonna stay below 30 or above 70, like may 9th to july 14th. Obviously in trending patterns you need a very good risk management system

    from may 9 to july 14 you had 5 stochastic lows below 30, yet only the last stochastic low proved to be the price low. that poses the danger if you base your entry on just the tops or bottoms of the stochastics patterns, which you can't predict in advance.

    but look at 28 may, you had a small bounce, if you look at 22 jan u will see why, it was a support level, so you had a $1 bounce, when stochastics was turning up a bit from a bottom.. so stochastics can be useful, but you should probably make an entry using price analysis
    unless you're quite good with just indicators but i don't know anybody who just uses indicators without price.

    looking at july 20 support became resistance and stochastics seemed to turn down too so you could have had a nice short there.
  9. KS96


    you have no exit strategy
  10. Suppose price values decrease to new lows then rally 10,000,000,000 %. You missed it.
    #10     Sep 20, 2008