maybe also decoupling tracking quality levels from tracking itself would be a good idea, splitting it into a maneuverability stat

Currently, we have three distinct characteristics affecting a missile's overall ability to track.

Firstly, there's the tracking type and value - "infrared tracking" 0.65, for example. This governs the missile's chance to retain its lock on a given target. Depending on the type, that's affected by different factors.

Next, there's the homing - this can take one of four values, and essentially governs how smart the missile's AI is.
homing 1 is currently, as far as I'm aware, essentially a half-implementation of what you're suggesting, as it stops a missile seeing targets behind it.
homing 2 just causes the missile to continually thrust and always try to turn to point toward the target.
homing 3 is similar, but the missile will stop thrusting if the target is behind it, allowing it to turn around more quickly.
homing 4 causes the missile to calculate a lead point to aim for based on the target's speed and heading.

Finally, there's the maneuverability of the missile, which comes from its acceleration, drag and turn.
The missile's top speed, for example, is given by acceleration divided by drag. For the same top speed, making both numbers higher (for example, 5 acceleration and 1 drag, rather than 0.5 and 0.1) makes the missile reach its top speed faster, and also lets it change heading more quickly, because it accelerates faster. The turn is the simplest number, simply governing how quickly the missile can rotate. Note, though, that if the acceleration is low and the turning is high, the missile may well be able to turn to face the target quickly, but it will take a while to actually change its direction of motion, so for a while it'll be pointing at the target but drifting in the direction it was facing before.