# seasonality and generalized additive model

I am reading a report which talks about seasonality. There is a chart showing the average returns for each month of the year. In the chart it appears the last 3 months of the year tend to be negative.

Then they say 'to check if this is due to the timing of various crises or bubbles we have fitted a generalized additive model (GAM) to the returns time series of each stock,

  Rij = Fi(t) + month(t) + Eit

please note i, j & t (when not in brackets) are subscripts


Fi(t) is an arbitrary but smooth function.

The two charts show the same pattern.

I have to be honest I have not come across GAM (or am unaware I have used it before). I do not understand the use of this GAM & how they have just picked some arbitrary function and this proves their point?

• Should be migrated to stats.stackexchange.com – Ric Oct 8 '14 at 10:39
• The application seems to be in finance and how the method is applied in finance and whether this is even correct. So this is quite domain specic and therefore I think it's better to keep it here. – Bob Jansen Oct 8 '14 at 13:32
• I can understand the logic in the question being moved to stats.stackexchange. However I agree with @BobJansen. – mHelpMe Oct 8 '14 at 14:42
• I don't mind having it here, but in my experience the OP gets better and quicker answers about GAM over there at SSE. – Ric Oct 9 '14 at 6:31