Similarity in .conf_hi and .conf_lo Differences for Forecast Data Points

[dmresearch15]

I am using modeltime_forecast with conf_interval = 0.95 and conf_method = "conformal_default”.

ex_ante_obj <-
calibration_tbl %>%
modeltime_forecast(
new_data = future_tbl,
actual_data = sub_train_data
)

In most cases, I observe that the difference between .conf_hi and .conf_lo is nearly the same for new_data. Could you help me understand why this occurs?