I can replicate on 1.5.0. I think that syntax will be changed with the new released. It’s already in master. You can do:

from statsforecast.utils import ConformalIntervals

# Create a list of models and instantiation parameters 
intervals = ConformalIntervals(h=24, n_windows=2)

mstl = MSTL(
    season_length=[24, 24 * 7], # seasonalities of the time series 
    trend_forecaster=AutoARIMA(prediction_intervals=intervals) # model used to forecast trend,
)
sf = StatsForecast(
    models=[mstl], # model used to fit each time series 
    freq='H', # frequency of the data
)
sf = sf.fit(df=df)
forecasts = sf.predict(h=24, level=[90])
forecasts.head()

The

prediction_intervals

on the

MSTL

class doesn’t work quite yet. You need to use it on the models (again, on the master branch)

MSTL does not have its own conf. intervals. ?