Stop when it agrees
The floor is boards of equal width. A needle lands with its middle anywhere and its direction anything, and it either lies across a joint (russet) or does not (green). If the needle is L long and the boards are d wide, it crosses with chance 2L / πd.
Buffon's game uses the needle he worked out in 1777, π/4 of a board, “almost three quarters”, so that the two players, one betting on a crossing and one against, have even chances. The trace is the share that crossed, and it should settle near one half.
Wolf uses Rudolf Wolf's proportions, 36 mm on 45 mm, and his five thousand throws, and reads the floor backwards as Laplace proposed: π ≈ 2Ln / dh after n throws and h crossings. He got 3.1596. Yours will be off by about as much, in one direction or the other.
Lazzarini's checks uses his needle, five sixths of a board, and throws in batches of 213. After each batch it asks one question: are the crossings an exact multiple of 113? When they are, the estimate is exactly 355/113, which is 3.1415929, and it stops there, the way his 1901 report probably did after sixteen batches. The ticks on the trace are the checks. Keep throwing afterwards shows what the next batch would have said.