Can someone explain why my Time Series (ARIMA, ETS) are flatlining and how to resolve?

[AkisM]

I'm attaching my workflow below.

Basically I have monthly sales data for 4 years. I split my dataset to first 2 years and next 2 years using a filter tool.

I try to compare ARIMA to ETS, but both models are flatlining. Can someone assist please?

[gawa]

@AkisM 

By changing configuration related to "seasonal" option, it shows some trend anyway, though I don't fully understand its mechanism😅 

[aatalai]

@AkisM an aside but I would suggest only using 6-9 months to test the model and the rest for training rather splitting it 2year and 2 years