Witt (1996) has shown that for an elliptical potential, the four images of a quadruply lensed quasar lie on a rectangular hyperbola that passes through the unlensed quasar position and the center of the potential as well. Wynne and Schechter (2018) have shown that, for the singular isothermal elliptical potential (SIEP), the four images also lie on an `amplitude' ellipse centered on the quasar position with axes parallel to the hyperbola's asymptotes. Witt's hyperbola arises from equating the directions of both sides of the lens equation. The amplitude ellipse derives from equating the magnitudes. One can model any four points as an SIEP in three steps. 1. Find the rectangular hyperbola that passes through the points. 2. Find the aligned ellipse that also passes through them. 3. Find the hyperbola with asymptotes parallel to those of the first that passes through the center of the ellipse and the pair of images closest to each other. The second hyperbola and the ellipse give an SIEP that predicts the positions of the two remaining images where the curves intersect. Pinning the model to the closest pair guarantees a four image model. Such models permit rapid discrimination between gravitationally lensed quasars and random quartets of stars.

We develop a robust method to model quadruply lensed quasars, relying heavily on the work of Witt (1996), who showed that for elliptical potentials, the four image positions, the source, and the lensing galaxy lie on a right hyperbola. For the singular isothermal elliptical potential, there exists a complementary ellipse centered on the source which also maps through the four images, with the same axis ratio as the potential but perpendicular to it. We first solve for Witt's hyperbola, reducing the allowable space of models to three dimensions. We then obtain the best fitting complementary ellipse. The simplest models of quadruple lenses require seven parameters to reproduce the observed image configurations, while the four positions give eight constraints. This leaves us one degree of freedom to use as a figure of merit. We applied our model to 29 known lenses, and include their figures of merit. We then modeled 100 random quartets. A selection criterion that sacrifices 20% of the known lenses can exclude 98% of the random quartets.