The current firmware version supports two FIR filters, namely peak and average. These are used in fast (oversampled) mode for display rendering (raw capture data used for export is not filtered). I also implemented a version of the sinc filter (see Nyquist sampling theorem and Whittaker-Shannon) which is a popular low pass IIR filter. The brute force implementation added about 75s (one minute and 15 seconds) per buffer cycle. A highly optimized version of this filter may approach 1/10 of that, but at 7.5s per cycle (which may be unrealistic) it is still not useful for practical purposes.
Most IIR filters will likely require processing beyond the capability of the Nano M3 Cortex as what we currently have is already a stretch at the maximum rate of 1MS/s. Another issue is that every single byte of RAM is used for the existing feature set and so adding new functionality is not possible unless some feature is also removed. One popular request has been for FFT (Fast Fourier Transforms) and at moderate (audio) frequencies, this is achievable, but not in combination with a rich general purpose feature set as we have now.
Those interested in filter/sampling theory can use the Nano capture capability, transfer the files to a PC and post-process the data using tools like Matlab.