Dear Statalists,

Please, I would really appreciate your help in my analysis.

I am working with hourly electricity prices in the Nordpool market. I have hourly electricity prices from 2007 to 2016 and I want to see the effect of a policy intervention in the market in November 2011.
Electricity prices are characterized by daily (more consumption around 11am and 4pm), weekly (less consumption on saturdays and sundays) and monthly (more consumption in winter) seasonality.
In order to check for the white noise I used, firs of all, two different techniques and I investigated what happened to the (logged) prices in a narrow window around the policy intervention.
I used the moving average with 96 hours (3 days, in order to take into account both hourly and weekly seasonality) and, by using a reference system price, I also regressed both the variables on the hourly and weekly dummies and then regressed the DV residuals on the IV residuals.
Both of these analyses showed that the volatility of prices decreased after the intervention.
Now, I am trying to model my data with ARMA process (time series is stationary).
However, it seems (and actually there is) a sinusoidal pattern that I do not know how to address.
According to my AC and PAC I should do something like ARMA(1, 120) which seems a little weird.
Could you please suggest me what I am missing?
Please, find attached the graphs from the AC and PAC.





Array Array

Thank you
Luisa Loiacono