I don’t know how WHR works in detail. But I imagine it to be a black box with output that is very much like that of an Elo rating system.
I think the black box does more iterations than a normal Elo rating system (perhaps somewhat similar to Glicko2?) and I think it also uses back propagation (updating past ratings), unlike Elo rating systems. But whatever the details of those added calculations, I assume that WHR rating differences still correspond to winning probabilities, very similar to Elo rating differences.
I can give empirical formulas to convert handicaps/rank differences to winning probabilities (or equivalently, Elo rating differences, and perhaps WHR rating differences if my above assumption is true). It comes from the historical data of the European Go Database (a database of 1 million live tournament games played in Europe since 1996). See http://goratings.eu/Home/About and Rating system issues
Note that my formulas will not match OGS handicap/rank-rating conversion formulas.