We propose a new method of estimation of the black hole masses in AGN based on the normalized excess variance , \sigma ^ { 2 } _ { nxs } .
We derive a relation between \sigma ^ { 2 } _ { nxs } , the length of the observation , T , the light curve bin size , \Delta t , and the black hole mass , assuming that ( i ) the power spectrum above the high frequency break , \nu _ { bf } , has a slope of -2 , ( ii ) the high frequency break scales with black hole mass , ( iii ) the power spectrum amplitude ( in frequency \times power space ) is universal and ( iv ) \sigma ^ { 2 } _ { nxs } is calculated from observations of length T < 1 / \nu _ { bf } .
Values of black hole masses in AGN obtained with this method are consistent with estimates based on other techniques such as reverberation mapping or the M _ { BH } -stellar velocity dispersion relation .
The method is formally equivalent to methods based on power spectrum scaling with mass but the use of \sigma ^ { 2 } _ { nxs } has the big advantage of being applicable to relatively low quality data .