AI
Sunday, January 03, 2016 11:38:26 AM
loanwolf, looked at your chart and analysis today.
i wanted to backtest your indicators against other days before replying
but too much happening this weekend.
have you tried this indicator config against other periods/fractals?
the signals definitely look promising for tracking thurs' chart.
Wm%R, aroon9, ma9 gave pretty optimal signals for that day.
my concern is how accurate some of the signals may be on choppier periods.
which is played out a bit in late thurs afternoon.
as a quick test, below is a chart with these indicators applied to recent SPY daily.
the Wm%R signals tracked the october rally pretty well
from up breach -80 to down breach -20.
aroon9 was less useful for october,
offering a late buy signal
that was only still useful because the october rally lasted so long (uncommon).
compare with aroon9 tracking in nov/dec.
ma13 was also useful for october, but too slow a signal in nov/dec.
ma9/ema9 were more accurate in october, and slightly better in nov, but useless in dec.
my general thought is that all of these signals will work
but their accurately depends on determining the optimal time period params for them.
aroon is am an easy example therein.
due to it simply measuring time since last low/high,
it's only useful is you use to correct time period for the chart in question.
too long period on choppy chart --> useless.
too short period --> too rapid signaling on brief chart hiccups.
eg: aroon5 crossings signal entry/exit of the october rally very well
but are very useless in nov/dec due to more rapid chart oscillations.
nutshell, these are great indicators if we can determine a way to reliably determine the underlying oscillation period of the chart.
which we might be able to guesstimate from STO patterns.
relying on prior trends is obviously unreliable as oscillation periods change pretty rapidly.
eg: oct vs nov vs dec.
let's keep trying things and talking about params.
as posted earlier, my goal is to develop composite indicators
that compile signals from a large number of individual indicators
to give a better sense of when tides are shifting.
this will obviously generate middling chaos most of the time.
but it may be useful for detecting setups when many stars are aligning
signaling reversal.
and there is wisdom in waiting to play big money until such setup arise.
eg: B.E.T's notion that trading less is more.
i wanted to backtest your indicators against other days before replying
but too much happening this weekend.
have you tried this indicator config against other periods/fractals?
the signals definitely look promising for tracking thurs' chart.
Wm%R, aroon9, ma9 gave pretty optimal signals for that day.
my concern is how accurate some of the signals may be on choppier periods.
which is played out a bit in late thurs afternoon.
as a quick test, below is a chart with these indicators applied to recent SPY daily.
the Wm%R signals tracked the october rally pretty well
from up breach -80 to down breach -20.
aroon9 was less useful for october,
offering a late buy signal
that was only still useful because the october rally lasted so long (uncommon).
compare with aroon9 tracking in nov/dec.
ma13 was also useful for october, but too slow a signal in nov/dec.
ma9/ema9 were more accurate in october, and slightly better in nov, but useless in dec.
my general thought is that all of these signals will work
but their accurately depends on determining the optimal time period params for them.
aroon is am an easy example therein.
due to it simply measuring time since last low/high,
it's only useful is you use to correct time period for the chart in question.
too long period on choppy chart --> useless.
too short period --> too rapid signaling on brief chart hiccups.
eg: aroon5 crossings signal entry/exit of the october rally very well
but are very useless in nov/dec due to more rapid chart oscillations.
nutshell, these are great indicators if we can determine a way to reliably determine the underlying oscillation period of the chart.
which we might be able to guesstimate from STO patterns.
relying on prior trends is obviously unreliable as oscillation periods change pretty rapidly.
eg: oct vs nov vs dec.
let's keep trying things and talking about params.
as posted earlier, my goal is to develop composite indicators
that compile signals from a large number of individual indicators
to give a better sense of when tides are shifting.
this will obviously generate middling chaos most of the time.
but it may be useful for detecting setups when many stars are aligning
signaling reversal.
and there is wisdom in waiting to play big money until such setup arise.
eg: B.E.T's notion that trading less is more.
99.99% of all pinks are scams. Best to assume the other 0.01% are as well.
