Suppose that a sequence of $${\varvec{n}}$$
n
cards, numbered 1 to $${\varvec{n}}$$
n
, is placed face up in random order. Let $${\varvec{k}}$$
k
be the number on the first card in the sequence. Then take the first $${\varvec{k}}$$
k
cards from the sequence, rearrange that subsequence of $${\varvec{k}}$$
k
cards in reverse order, and return them to the original sequence. Repeat this prefix reversal until the number on the first card in the sequence becomes 1. This is a one-player card game called Topswops. The computational complexity of Topswops has not been thoroughly investigated. For example, letting $${\varvec{f}}({\varvec{n}})$$
f
(
n
)
denote the maximum number of prefix reversals for Topswops with $${\varvec{n}}$$
n
cards, values of $${\varvec{f}}({\varvec{n}})$$
f
(
n
)
for $${\varvec{n}}\ge 20$$
n
≥
20
remain unknown. In general, there is no known efficient algorithm for finding an initial sequence of $${\varvec{n}}$$
n
cards that requires exactly $$\ell $$
ℓ
prefix reversals for any integers $${\varvec{n}}$$
n
and $${\varvec{\ell }}$$
ℓ
. In this paper, using a deck of cards, we propose a physical zero-knowledge proof protocol that allows a prover to convince a verifier that the prover knows an initial sequence of $${\varvec{n}}$$
n
cards that requires $${\varvec{\ell }}$$
ℓ
prefix reversals without leaking knowledge of that sequence. We also deal with Botdrops, a variant of Topswops.