SGP4 is the standard algorithm that turns a TLE into a satellite's predicted position at any moment, modelling Earth's oblateness, atmospheric drag and lunisolar effects. Nearly every tracking app and screening system runs it — with typical kilometre-scale accuracy that degrades as the TLE ages.
Understanding SGP4
Why a 1970s model still runs the sky
SGP4 survives on three virtues: it is fast enough to propagate tens of thousands of objects every second on modest hardware; it is the format's twin, since the global TLE pipeline fits to it; and it is a stable, published standard — the 1980 Spacetrack Report No. 3 and its 2006 revision gave every implementer the same equations. The entire public tracking ecosystem, this site included, stands on that common core.
Where its limits bite
The drag model is SGP4's soft spot: a single B* coefficient cannot capture an atmosphere that breathes with solar storms, so low-perigee objects and re-entry candidates carry the largest errors. Manoeuvring satellites break it differently — a burn invalidates the fitted elements until a fresh TLE appears, which is why newly manoeuvred constellation satellites briefly appear "off track" in public data.