OK, I saw that not the grid sorts the rows. The dataset behind that does it, so the OverIndex is the dataset row index too.
But. Is there another type of drop? Now there is a green line between the destination grid's rows. In this case if I drop something on a row then it can give me 2 indexes: the right row index and the next row index depends on the green line's position. In my case I don't want to move/copy rows from one grid to another, I want to "assign" rows from the source grid to one row of the destination grid. I don't need the green line and I want one overindex independently of the drop position on one destination row.
Is there a solution to this?