A transit happens when a planet's orbit carries it directly between its star and a distant observer, briefly blocking a sliver of starlight. The planet itself is far too faint and small to see directly, so this dip in brightness is often the only evidence it exists at all.
The depth of the dip is set by how much of the star's disk the planet covers, scaling with the square of the radius ratio, a planet twice as wide blocks four times as much light. The slanted edges of the dip aren't instant either: as the planet slides onto and off the stellar disk, the brightness ramps down and back up, with the flat bottom marking the period when it's fully superimposed on the star.
For a transit to be visible from Earth, the planet's orbital plane has to be tilted almost exactly edge-on to our line of sight. Most planetary systems aren't aligned this precisely, which is why the transit method only catches a small slice of the exoplanets actually out there, but because it repeats every orbit, watching several identical dips in a row also reveals the planet's orbital period.
Real exoplanet orbits are rarely perfect circles. Orbital eccentricity stretches the ellipse, a value of 0 is circular, while values approaching 1 are highly elongated. An eccentric planet moves fastest at periapsis (closest approach) and slowest at apoapsis (furthest point), following Kepler's second law. This means transit timing and duration can vary significantly with eccentricity, and the dip in the light curve may look asymmetric if the planet is accelerating or decelerating through the transit.
