markus it's not super elegant, but what do you think about a util like:
discontinuous_series_helper(lambda series: ta.sma(series, 20), close_stocks)
# where close_stocks is, for example, `[120, 150, 145, 108, 104, nan, nan]`
# the util's procedure is as follows
## if `close_stocks[-1]` is `nan`, it retrun `nan` for the current value
## if `close_stocks[-1]` is not `nan`, call the lamda fnc with `close_stocks` sans all the `nan` values (ie filter them out)
# do this recursively if a series of past values is desired
This would make it so that TA fncs for stocks after closing time would return a current value of nan, but for time periods where there are candle values, the indicator value(s) would be as expected. This seems like OK behavior to me. Lemme know if i'm missing something though 😃