Disliked{quote} These indicators look perfect in hindsight but it cannot be traded in real time, since they "readjusting" with every tick of data?Ignored
The point I wished to make is that indicators are rendered using historical data. Indicators are useful for looking for patterns to get ideas for how to trade. They might be useful for trading, but only if they don't repaint. And, with respect to the issue at hand, the goal is NOT to make the Heiken Ashi indicator look a certain way. As an indicator, we can make it look a variety of different ways -- but as to which one is best for trading no one knows.
I've previously posted why this Heiken Ashi looks different than others -- it starts it's computation at a different index in the data. There is no right or wrong index to begin computing from. But beginning at different indices will yield slightly different HA transitions. I've chosen to use the most recent 500 bars because I'm concerned about CPU performance. Furthermore, my intuition tells me that using the most recent data probably yields a more relevant result than using 1000's of bars of stale data that would skew the results. And, finally, we can see not only with our own eyes but also from results that people have shared that the current Heiken Ashi transitions create very nice Entry signals. So -- there is no problem with Entry signals.
Two problems have recently come to light -- both related to repainting:
(1) Missed midline crossings -- which I believe are due to repainting caused by a nearby large spike. As I wrote previously, I don't think midline crossings are missed -- rather, a subsequent large spike pulls the midline up or down such that subsequent rendering by the indicator makes it look like a midline crossing was skipped. However, at the time where it now looks like there was a crossing there really wasn't one. So, just as rendering the ZigZag indicator makes it look like there are good Entries and Exits, during real-time they are not apparent.
(2) We may need another Exit protection to guard against these extreme prices -- especially if trading M15. It is our impression this problem is isolated to LTFs. We havn't yet encountered it on H1, H4 or D1.
We are currently looking at the possibility of adding another Exit indicator. However, as with everything, there are trade-offs. An additional Exit indicator could protect against spikes, but it will also probably cause some winning trades to exit prematurely.
2