Run reactivity — how hard the opponent model chases positional runs it has actually observed (0 = pure ADP, 2 = strongly herd-following). Bench weight — credit given to a player who does not crack your starting lineup. Bench weight now scales an option value: a bench player is priced on his chance of beating the man ahead of him, so high-spread players are worth more once your starters are set. Starter injury risk lowers that bar. Workload tilt is how much your spreadsheet's More/Less column moves a projection. Lookahead — how many of your own future picks the search plans over (1 = greedy, 3 = default). Sims — Monte Carlo opponent trajectories sampled per recommendation.