04-30-2018, 05:27 AM
(This post was last modified: 04-30-2018, 10:18 PM by susitronix.)
i have to thank. thanks Mike
its amazing what i have learnd with gekko/java script.
we will build a little strategie but, that tryes to keep up with the big boys.
found that its pretty difficult with the RSI since in the sidways market the RSI falls back to the middle and neglect to goShort,
>>>therefore keep that bag when the trend goes down (wedge analysis, crack the resistance...).
RSI seems usefull for the Long signal (sometimes!)
since my new Trendline indicator is not finished yet, a intresting idea has derived:
a simple trailing loss.
tryed with RSI-long and trailling-stop short, but it instantly goes long again after the short because the loss does best with -3%.
thats where the RSI has its lower threshold.
did not tryed hard only one night.
it seems that this parameter must be dynamicly adjusted. imagine:
the long level creates a lower and a upper zone.
the lower zone would hardwired trigger at -1.2% inorder to have enough headroom for the nois.
but if it goes above long level, the sample+hold would charge up. the upper zone from 0% to 1% above long would have -2% trailing loss.
then from 1%-2% above long it would go to -3% trailing loss.
>>>but then at some point it should switch to -0.5% for to goShort..???
I guess this cant work that way because of the the market signal is very dynamique.
so we would try to make trend analysis, that indicates the trend.angle and the trend.amplitude.
thatway we may teach our gekko to set the trailing stop to -0.5% or so.
Since it goes so fast sometimes, a reactive mecanism is better than nothing but a pro-active goodness would be more in-time.
Also if there is a crack in the support, then it should not really try to trade.
Its seems hard to catch the harsh down/moon trends where the timing would tradeoff aginst smoothness/persistance.
a trend line indicator works better there. it would sample the highs and lows and draw a vector line based on the difference low-low.
this blocks the higher harmonic by a sort of integrating mechanism while live from a shorter averaging in order to be smooth,
than regular indicators.
offcourse it very much blocks the frequencys below the macro.trend frequency.
and if there is a crack in the resistance then i could reset and preload the newest difference,
into the averaging cells (support/resistance vectors) for a faster trend reversal
its amazing what i have learnd with gekko/java script.
we will build a little strategie but, that tryes to keep up with the big boys.
found that its pretty difficult with the RSI since in the sidways market the RSI falls back to the middle and neglect to goShort,
>>>therefore keep that bag when the trend goes down (wedge analysis, crack the resistance...).
RSI seems usefull for the Long signal (sometimes!)
since my new Trendline indicator is not finished yet, a intresting idea has derived:
a simple trailing loss.
tryed with RSI-long and trailling-stop short, but it instantly goes long again after the short because the loss does best with -3%.
thats where the RSI has its lower threshold.
did not tryed hard only one night.
it seems that this parameter must be dynamicly adjusted. imagine:
the long level creates a lower and a upper zone.
the lower zone would hardwired trigger at -1.2% inorder to have enough headroom for the nois.
but if it goes above long level, the sample+hold would charge up. the upper zone from 0% to 1% above long would have -2% trailing loss.
then from 1%-2% above long it would go to -3% trailing loss.
>>>but then at some point it should switch to -0.5% for to goShort..???
I guess this cant work that way because of the the market signal is very dynamique.
so we would try to make trend analysis, that indicates the trend.angle and the trend.amplitude.
thatway we may teach our gekko to set the trailing stop to -0.5% or so.
Since it goes so fast sometimes, a reactive mecanism is better than nothing but a pro-active goodness would be more in-time.
Also if there is a crack in the support, then it should not really try to trade.
Its seems hard to catch the harsh down/moon trends where the timing would tradeoff aginst smoothness/persistance.
a trend line indicator works better there. it would sample the highs and lows and draw a vector line based on the difference low-low.
this blocks the higher harmonic by a sort of integrating mechanism while live from a shorter averaging in order to be smooth,
than regular indicators.
offcourse it very much blocks the frequencys below the macro.trend frequency.
and if there is a crack in the resistance then i could reset and preload the newest difference,
into the averaging cells (support/resistance vectors) for a faster trend reversal