Not sure what are the data and source of data? B3=53, is this sale? what is the rank BN?
Maybe you look at sales and compare to end of year votes? In this case, you will never have full "agreement" between the model and the rank
Yearly rank is done once a year and depend on people mind at the end of the year.
I assume that like doing a regression,(in a regression model the estimate Y and Y are not always equal) , your model will not predict exactly the yearly rank. And if you find model that will predict the numbers it would be probably over fit and won't predict other data.