Just a guess. The index never converged so the RA play was the best. There are sims running for analyzing a new game I am going to attack. The playing sims are starting to converge but a lot of the indices are slow to converge and some will likely never converge. They bounce back and forth between the 2 decisions as the TC changes. Either no index is ever converged upon and/or the correlation to the play is so poor that the count being used isn't worthy of being a barometer for the decision. Either way the RA play becomes the preferred option for the grey area where the correct play as viewed by the count is not converging for a range of TC. If the decision is always the same except for the grey area only one play is indicated but you can use the RA play for the grey area if the decision made otherwise is a double or split.