does ahg use heiken ashi? i never thought they were a good idea cuz you're not seeing actual price data. better to use a fast, smooth MA over real candles for seeing trend. by the time a HA trend reverses, not only is it usually one bar too late, but price is farther away from you than you think because you're not seeing actual ohlc data. please explain what i am missing if you are reading this and use HA bars profitably. i just cant believe the price action "all indicators lag" crew would use HA, because it lags.
Early on, as an aid in "staying with the trend" some AHG folks looked at HA as a possible method. I think it has generally been abandoned because of the reasons you mentioned. In the end, the trend will not change untill a signifcant H or L pivot has a swing above/below it. Pure PA. You can estimate it with a trendline, a moving average or HA bars, or any number of derived inicators. FWIW
HH-LL code for Amibroker: Code: [color=blue] //SUPPORT & RESISTANCE LEVELS //Graham Kavanagh 9 Oct 2003 #49421 //Find turning points top = H==HHV(H,3) AND H>=Ref(H,1) AND H>=Ref(H,2); bot = L==LLV(L,3) AND L<=Ref(L,1) AND L<=Ref(L,2); topH = ValueWhen(top,H); botL = ValueWhen(bot,L); //Indicator section GraphXSpace=5; col = IIf( Close > Ref( Close, -1 ), colorGreen, colorRed ); Plot( Close, "Price", colorWhite, styleCandle ); Plot( topH, "res", colorRed, styleDots+styleNoLine ); Plot( botL, "sup", colorBrightGreen, styleDots+styleNoLine ); Title = Name() + ", " + Date() + ", Support & Resistance Levels Support = " + topH + ", Resistance = " + botL ; [/color]
kiwi, ditto here, I'm still working on it, but so far my automated strategies based on price action are only VERY marginally profitable. I'm not sure yet if it's possible. Been sim trading AHG discretionary and it's SO easy to "read" what's going on in real time, and SO hard translating that into code. Keep at it though, in principle if it can be done it can be programmed...
Never say never ... but almost impossible to code PA. Someone, somewhere, sometime, must have done this --- let us know. I'd like to backtest anythng I trade but I am afraid doing that with PA - the PA that is dicussed at length in these forums - is extremley difficult. There must be definite clear-cut algorithms for S/R etc with parameters that can be optimized and backtested. Anyone has done this with PA?
use Donchial study <img src=http://www.elitetrader.com/vb/attachment.php?s=&postid=2200085 width=800> click on link to enlarge image 256
Anything that can be measured and quantified, it can be programmed. Having said that... we should also recognize the statistical reality... If we look at a 3 bar combination of OHLC... we can arrive at more than 47,000 combinations of PA patterns.