Disliked... Pips4life, great work you have done...
I stumbled on the Fish indicators which seemed good for TSS, but if it is found to repaint then into the dustbin it goes.
...
I am not trying to defend Fish or TSS, far from it. If something does not work, toss it and move on to something hopefully better.
...
I have noticed that Fish operates on both bar1 and bar0 the current bar, and have also noticed a disappearing arrow (at bar 0) even on H1 TF. That's to be expected.
...Ignored
Hi Snobol1,
Thanks but I don't deserve much credit. Others, like Chandra, are doing a lot more actual testing than I am. I'll remind you that I'm still a beginner at Forex and MT4, so I still have plenty to learn. If I give suggestions that are just flat out wrong, please don't hesitate to correct me.
Last week I made posts about a couple of topics.
The first was to say the current indicator has some problems with wrong-way arrows. If you display the raw FX_FISH_2MA data, you can tell these are bad arrows because they're occurring at times where no actual zero-crossing occur. This bug should be fixed, especially since it may interfere with an EA. Getting a wrong-way arrow would be quite bad for automatic trading.
My second topic was about my frustration with not being able to redraw or change timeframes back-and-forth because it moves the arrows back in time, or if they were "bad" arrows, it eliminated them entirely. I still wish the behavior of this indicator were different, and I remain hopeful it can be improved but the programming is beyond me.
Having said that, I did run my twelve H4 charts from Tues-Fri, and then I reviewed the arrows. I didn't redraw, so I think the arrows were still in the position they had occurred in real-time. If I had traded those arrows, even some that were many bars old, I saw that two of the charts would have probably been bad results with some losses, but the other 10 could easily have resulted in gains of: +20, +45, +50, +60, +80, +100, +100, +100, +100, and finally +160 (USDCAD, which has been on an almost relently downtrend for weeks). I asked myself, WHAT was I complaining about??!
This week, I've been demo trading the H4 arrows, and though they don't happen very often, I've seen more success than failures. So, I do think this FISH indicator deserves more time to fully check it out.
Regarding your comment about throwing repainting indicators into the dustbin, I'm all in favor of that, HOWEVER, I don't think FISH is exactly a repainting indicator.
Maybe I don't understand the definition, but what I *think* repainting means is that without doing a redraw, an indicator will change the chart (auto-redraw) and move even very old arrows/lines/histograms around to perfectly fit the data. This would indeed be a worthless indicator if that were the case. Other than Bar0 arrows which may blink on/off/on/off, which is fine and to be expected, I have never seen older data change BY ITSELF. Rather, FISH is what I'll call a perfect-hindsight indicator, which is that if you start it fresh, or if you do a complete Redraw or change Timeframes (e.g. H4->H1->H4), then indeed, the arrows do move, sometimes substantially. I saw one example of an up-arrow 4 bars back which upon a redraw moved to 16 bars back ( a change of -12 bars!!). This was because those -12 bars were relatively flat.
Let's give FISH some more time. I'm also in favor of checking out other indicators in parallel as well. I'd also like to know if the EA has been put to the test and what are the results. Can anyone post the EA so more people can try it? (Does the wrong-way arrow bug interfere with the EA? If so, that bug must be fixed).
It seems to me that the Fisher transform may well have merit, but the first indicators used methods of perfect-hindsight to adjust the history data. Because of this, it may well be that people abandoned those past indicators. Maybe, however, that was throwing the proverbial baby out with the bathwater. The transform may well be good, but it just needs a fresh attempt at recoding it without the perfect-hindsight aspect.
If I were to describe a perfect TSS indicator, it would do:
1. Upon a fresh start or redraw, the arrows occur at exactly the same place they would have for real-time data. No arrows get shifted backwards in time, nor do they disappear if they turn out "bad".
2. Though not required, it would be helpful if a different marker could indicator that during a particular bar, a blinking on/off arrow had occurred but by bar close, it was gone.
3. An arrow is place only above/below Bar0 (the current bar) which formed that arrow. An arrow is never placed at Bar1 or farther back in time.
4. All arrows are determined by history price action and Bar0 action. As you process a stream of historical data, you NEVER should consider "future" price action to determine whether to draw an arrow. As this is exactly what true real-time action does, I see no reason to do anything different for historical data. For every bar, since what you know if OHLC data but not the actual order, then the code may consider two possibilities as to whether an arrow would have been formed. Consider open->high->low->close, and consider also open->low->high->close. If either of these were to cause an arrow, indicate this on the chart, either as an arrow, or as another kind of symbol (a triangle, or different style arrow, whatever...).
Bottom line: NEVER consider future data when calculating historical arrows.
5. Fix the current FISH bug with wrong-way arrows.
We did have some programmers following this thread in the past, and if any are still reading this, are you interested to work with this FISH indicator to improve it?
Thanks for reading this long post...
Pips4life