I am trying to estimate the background from fitting the ET bin by ET bin signed pT balance distribution.  The "rige" in the 2D , sPtBalance vs. ET, plot is expected to be dominated by the W candidates. The QCD events should not give such an clear "x=y" rige, and they are expected to have a relatively smooth distrubution. So we can estimate the background under the salient signal peak. 

Below plot shows the signed pT balance distribution projected from the 2D plots shown above.  By fitting it with a combined function, Gaussian background and lorenzian signal, we can get the yields from this ET bin.  With this method, we can get all the yields for the ET bins of interest.  Detail plots for every fitting can be found at below attachment. 

Below are the rough results from this method, for W+ and W- eta bin by eta bin respectively.  The electroweak backgrounds should be under the "x=y" rige due the miss detected neutinos or electrons.