Solution:
First of all, the game finishes because the number on the calculator always decreases. By picking m% of a positive integer n, players get the number
100m⋅n=2252m⋅n
We see that at least one of the powers of 2 and 5 that divide n decreases after one move, as m is not allowed to be 100, or a multiple of it. These prime divisors of n are the only ones that can decrease, so we conclude that all the other prime factors of n are not important for this game. Therefore, it is enough to consider numbers of the form n=2k5ℓ where k,ℓ∈N0, and to draw conclusions from these numbers.
We will describe all possible changes of k and ℓ in one move. Since 53>100, then ℓ cannot increase, so all possible changes are from ℓ to ℓ+b, where b∈{0,−1,−2}. For k, we note that 26=64 is the biggest power of 2 less than 100, so k can be changed to k+a, where a∈{−2,−1,0,1,2,3,4}. But the changes of k and ℓ are not independent. For example, if ℓ stays the same, then m has to be divisible by 25, giving only two possibilities for a change (k,ℓ)→(k−2,ℓ), when m=25 or m=75, or (k,ℓ)→(k−1,ℓ), when m=50. Similarly, if ℓ decreases by 1, then m is divisible exactly by 5 and then the different changes are given by (k,ℓ)→(k+a,ℓ−1), where a∈{−2,−1,0,1,2}, depending on the power of 2 that divides m and it can be from 20 to 24. If ℓ decreases by 2, then m is not divisible by 5, so it is enough to consider when m is a power of two, giving changes (k,ℓ)→(k+a,ℓ−2), where a∈{−2,−1,0,1,2,3,4}.
We have translated the starting game into another game with changing (the starting pair of non-negative integers) (k,ℓ) by moves described above and the player who cannot make the move loses, i.e. the player who manages to play the move (k,ℓ)→(0,0) wins. We claim that the second player wins if and only if 3∣k and 3∣ℓ.
We notice that all moves have their inverse modulo 3, namely after the move (k,ℓ)→(k+a,ℓ+b), the other player plays (k+a,ℓ+b)→(k+a+c,ℓ+b+d), where
(c,d)∈{(0,−1),(0,−2),(−1,0),(−1,−1),(−1,−2),(−2,0),(−2,−1),(−2,−2)}
is chosen such that 3∣a+c and 3∣b+d. Such (c,d) can be chosen as all possible residues different from (0,0) modulo 3 are contained in the set above and there is no move that keeps k and ℓ the same modulo 3. If the starting numbers (k,ℓ) are divisible by 3, then after the move of the first player at least one of k and ℓ will not be divisible by 3, and then the second player will play the move so that k and ℓ become divisible by 3 again. In this way, the first player can never finish the game, so the second player wins. In all other cases, the first player will make such a move to make k and ℓ divisible by 3 and then he becomes the second player in the game, and by previous reasoning, wins.
The remaining part of the problem is to compute the number of positive integers n⩽2019 which are winning for the second player. Those are the n which are divisible by exactly 23k53ℓ, k,ℓ∈N0. Here, exact divisibility by 23k53ℓ in this context means that 23k∥n and 53ℓ∥n, even for ℓ=0, or k=0. For example, if we say that n is exactly divisible by 8, it means that 8∣n,16∤n and 5∤n. We start by noting that for each ten consecutive numbers, exactly four of them are coprime to 10. Then we find the desired amount by dividing 2019 by numbers 23k53ℓ which are less than 2019, and then computing the number of numbers no bigger than ⌊23k53ℓ2019⌋ which are coprime to 10.
First, there are 4⋅201+4=808 numbers (out of positive integers n⩽2019) coprime to 10. Then, there are ⌊82019⌋=252 numbers divisible by 8, and 25⋅4+1=101 among them are exactly divisible by 8. There are ⌊642019⌋=31 numbers divisible by 64, giving 3⋅4+1=13 divisible exactly by 64. And there are two numbers, 512 and 3⋅512, which are divisible by exactly 512. Similarly, there are ⌊1252019⌋=16 numbers divisible by 125, implying that 4+2=6 of them are exactly divisible by 125. Finally, there is only one number divisible by exactly 1000, and this is 1000 itself. All other numbers that are divisible by exactly 23k53ℓ are greater than 2019. So, we obtain that 808+101+13+2+6+1=931 numbers not bigger than 2019 are winning for the statistician.