Hi,

If you shift a SMA of 100 periods half a cycle to the left (-50), you get a great picture, which other type of MA comes close to this curve?

I've tested a lot of things already and it seems that a JMA comes the closest. What do you guys think?

__________________
tensigma.blogspot.com

I think it would look like a 100 SMA shifted (-50) to the left " sorry just kidding"

I don't know really, but thanks for the observation, you give me some work to do.

__________________

I know going back in time is cheating on yourself, but it's not an entire trading system, it's just a simple moving average. Do you think a neural net could do this? Or some other "exotic" solution, or would it be more simple to do?

__________________
tensigma.blogspot.com

I know how the MAVG is calculated but I have never considered the shift , and I don't know why its there, so I google for the shift reason and I found this :

"Since a moving average looks back at prior measurements, it lags the current trend. You can shift the moving average backward in time to cancel this lag by entering the number of days of displacement in the Shift cell. This allows you to compare the shape of the trend curve found by various moving averages with the original trend."

So this gave me new ideas to test.

__________________

İs this any good?

A hybrid ARIMA and support vector machines model in stock price forecasting.(autoregressive integrated moving average) - Omega | Encyclopedia.com

Hi,
do you have a pdf link? I don't have ScienceDirect acces.
thx

__________________
tensigma.blogspot.com

Interesting shifted MA

Have you made any progress on this back-shifted MA?
I must say that I have tried many combinations of existing indicators to approximate the shape, but obviously I have failed so far.
I am now turning my attention to GMDH (Group Method of Data Handling (GMDH) for data mining, forecasting algorithms optimization, fuzzy models analysis, statistical learning networks and modeling software systems).

__________________
tensigma.blogspot.com