We propose a novel automatic shrinkage tuning technique for the adaptive proximal forward-backward splitting (APFBS) algorithm. The shrinkage tuning aims to choose an appropriate value of the shrinkage parameter and achieve minimal system mismatch as possible. The system mismatch is approximated based on time-averaged second-order statistics. Numerical examples show that the proposed method achieves performance fairly close to that with a manually chosen shrinkage parameter for colored input signals at some signal to noise ratio (SNR).