Methods of real time/online time series smoothing/filtering/noise removal that have low lag ( or no lag if possible )

Question

I have a time series data that's generated by a random function and streamed over to my other script. This script has a task of smoothing/filtering/removing noise from this streamed data in real time. I tried utilizing different algorithms and models like:

• Moving Average ( and different types of moving averages like least square, hull etc... )
• Savitzky-Golay method
• Whittaker-Eilers method
• Heat equation method
• Perona-Malik PDE
• Kalman filter
• 1 euro filter
• Low pass filter

And they all introduced a pretty good amount of lag during real-time/online streaming. Now, I understand that it may not be possible to remove lag further than I already did ( after all things I tried ) but I still have two more ideas that might make some sense:

• Incremental curve fitting on every new data point but with constraint that function must be the same as it with older data points ( I found some topics on this that correlate this with Kalman filter but I am not sure how to implement something like this )
• Estimation of moving averages ( I found out that this is maybe a topic that would work since I already tested some things on least square moving average estimation and got very low lag based on https://figshare.com/articles/preprint/A_NEW_LOW-PASS_FIR_FILTER_FOR_SIGNAL_PROCESSING/9762863 )

Now, keep in mind that I am just starting my journey into whole signal processing and causal filter waters and I already understand that zero lag filters is not possible because that's non-causal signal characteristic, I still would like to push this to it's absolute limits and try to find a filter or set of filters that would achieve this task reasonably well so that's why I am interested in various different ideas.

I already asked a similar question over at Signal processing: https://dsp.stackexchange.com/questions/95894/real-time-smoothing-using-savitzky-golay-method-or-any-similar-method-with-low-p