If a jump can be plotted such that a ship arrives at an "ideal" arrival point to get to the target planet in the shortest time, could a jump arrival point be predicted given an estimate of when the target left the prior system? Astrogation in reverse as it were.
In theory, yes; in practice, probably no.
A competent astrogator could look at the relative position and velocity of the origin world to the destination world, and assuming a full-thrust outbound run with no midpoint turnover from the origin world to the Jump point, and likewise a full-retrothrust run inbound to the destination world, for a given starship's M-drive rating, the geometry should be pretty straightforward to reverse-engineer and solve for time and place.
But there are two problems.
First, you have to deal with the whole "variable Jump duration" angle. Planets can move a great distance, especially with respect to each other, in a >32-hour window of uncertainty.
(For this reason, I have always contended that the duration of a Jump must be predictable from the Jump calculations
before the Jump begins, or else the margin of error is ponderously large otherwise and a starship can arrive quite far away from its intended destination or perhaps nearly on top of it; canon, of course, ignores this and lets Jump duration be a surprise only to be revealed once in Jumpspace, and perhaps then only with a few moments' notice once precipitation out of Jumpspace becomes immanent.)
So, if the position of the destination world can be reliably anticipated when calculating a Jump, it should be reverse-engineer-able from the arriving starship's position and vector (taking into account not only the two worlds' relative motion, but even things such any relative differences in planetary diameters and therefore specific differences in travel time out from one and in toward the other. But if it
cannot be pre-calculated, then no, you probably cannot reliably predict the exit time, place, and vector within a limit-of-military-sensor-range envelope. (Terra in its orbit, for example, shifts its position relative to Sol by ~30km/s.)
This is the easy part, as I see it. The hard part comes next.
Second, given that messages from one system to another cannot travel faster than the speed of Jump, how is a camping fleet of would-be ambushers going to know where to expect the arriving ship to be arriving
from?
If you can make a good guess (an Xboat or a Subbie or something like that on a
scheduled run), you can -- within the constraints of the first issue above -- lay your ambush (or, more charitably in the case of an Xboat Tender, set up for a quick rendezvous and capture/refueling) fairly reliably.
On the other hand, this might work only so long as an enemy remains unaware of your hunting tactics. Once alerted to your ploy, it is a trivial matter to move to less than "best speed" courses, and start Jumping in all over the place with whatever vector might be needed in the new geometry.
tl;dr: It depends on how fussy and in what way you are about variable Jump durations, and, that constraint still permitting, how benign/hostile you want the "welcome mat" practice to be in the long run.