This is a classic probability puzzle, whose result is counterintuitive.
For each $0<i< L$, let $p_i$ be the probability that $i$ is the last non-visited point. I claim that for any $1<i<L-1$, we have
$$
p_i=\frac12p_{i-1}+\frac12p_{i+1}\tag{1}
$$
To see this, suppose your first step is in the clockwise direction. Now, point $i$ is $i-1$ steps clockwise of you, so conditioned on this first step, the probability of seeing $i$ last is the same as the probability of seeing $i-1$ last. The other summand comes from conditioning on an anti-clockwise step.
The equation $(1)$ implies that in the list
$$
p_1,p_2,p_3,\dots,p_{L-1},
$$
each entry is the average of its neighbors, implying this list is an arithmetic progression. Furthermore, the symmetry of this problem implies $p_1=p_{L-1}$. Therefore, the arithmetic progression is constant, so each $p_i=\frac1{L-1}$; that is, no particular point is more likely to be visited last!