First of all, nice data and thanks for sharing. But I don't quite understand your last point. You are not mentioning anything like LMR or History Pruning, so I see no reason why the move ordering should affect the value of a node (just the node count). So why do you reckon null move success-rate would change?MattieShoes wrote:Good move ordering is necessary for null move pruning to operate effeciently. The faster you get alpha up, the more positions will fail high on depth+1.
One answer I would have is that when you prune in Quies non MVV/LVA captures, the prunings are actually unsound and thus would cause the nullmove to fail if weren't pruned.