In ages past, every turret had a different interpretation of their Turret Tracking/Accuracy Score trackng because they were relative to different “Optimal Signature Radii”. So a turret with a tracking of 300 might have better tracking than a different turret with a tracking of 500 simply because they were relative to different optimal signature radii. This made comparing turret tracking numbers tedious until, some years ago, CCP released a new system where they normalized all tracking numbers relative to the same optimal signature radii, thus making them directly comparable to one another: henceforth, 500 is strictly better tracking than 300 in all instances. (The actual tracking ability was unchanged, only the numerical representation of their tracking ability changed.)
Thing is… when they converted numbers to the new system, they forgot to do this for drones.
Example: the base stats of a Hobgoblin I indicate its OLD tracking score is 1.81; if you had compared this in-game to, say, a Light Neutron Blaster I’s tracking score of 379.8, you might think it to be grossly inferior and question how they manage to maintain even a tenth of their accuracy if their score is that bad.
In truth, the accuracy score, using the NEW system, is actually 2904 - which is 7.64x better than the turret.
Fitting tools like pyfa have the courtesy of converting to the new score for us, but both as a matter of principle and for in-game comparisons that inevitably come up by both those who don’t use fitting tools as often as they should and even by the occasional use by those who do (esp using Compare tool to compare modules of “different series”, eg. hybrids vs projectiles), CCP should finish the job and convert the tracking scores of drones to the new system released many years ago so that all tracking numbers (not just turrets) use the new system, and the old system ceases to exist entirely.
(Apologies to 3rd party developers who would need to remove the soon-to-be obsoleted conversion code. Less is more! Less is more! 
)