No. I don't know of any simple formula relating variance with the order statistics of selecting the max, such as in the worst-case scenario of picking from only

No. I don't know of any simple formula relating variance with the order statistics of selecting the max, such as in the worst-case scenario of picking from only 2. (In any case, the variance I used is 40%, then halved for relatedness.) I do have an open StatsExchange question on this: https://stats.stackexchange.com/questions/427292/probability-of-selecting-maximum-in-bivariate-correlated-order-statistics https://stats.stackexchange.com/questions/427292/probability... But to calculate the probability, I just Monte Carlo it.