# Oscillatory time-series forecasting

I was wondering if this mean(160)-reverting/oscillatory time series "SUM" can be considered chaotic & forecastable to some extend short-term?

http://sg.myfreepost.com/sgTOTO_analysispower.php?draws=60&fn0=Sum

Are there time-series forecasting methods that give a broader "range" prediction structure rather than "exact/precise" future numbers (e.g, neural network which can be totally off since it is just trying to curve-fit)? Is fuzzy logic method able to give a prediction interval for the immediate next period in a discrete time series?

I would appreciate if anyone can provide some insight to this case study.

-
 How is fuzzy logic applied to time series forecasting in general? Is anyone familiar with ANFIS? – Shelagh Oct 10 '12 at 20:45 mathworks.com/products/fuzzy-logic/examples.html?file=/products/… – Shelagh Oct 10 '12 at 20:50 I don't think I completely got your question. But Gaussian Processes [1] give a prediction performance estimate through the covariance matrix. [1] gaussianprocess.org – anabelli Oct 15 '12 at 19:47

http://justwebware.com/uklotto/uklotto.html

Test run quickpick

Test run 1,2,3,4,5,6

Test run (single digit,teens,twenties,twenties,thirties,forties)

1000 times or more each cycle for as many cycles as you want.

Seems like the \$returns are indeed better for the third option-middle SUM value or max kurtosis zone.

-