Using cumulative forecast error as the forecasting metric, naive forecasting is the best method in this case $($T$/$F$)$

Using sum squared error as the forecasting metric, exponential smoothing forecasting is the best method in this case $($T$/$F$)$

$$