[Keysh]

The "batting average" is a bit higher than that. For example, measurements of the proper motion (motion across the sky) of Sirius led to the prediction in 1844 that it was in an orbit with an (observed) faint or dark companion; the latter (the white dwarf Sirius B) was not directly observed until 1862, when better telescopes were available.

One could also argue that detections of planets from spectroscopic observations of stars is another example. The first observations of transiting exoplanets -- where the planet blocks some of the light of the star -- were actually cases where the existence of the planet had been previously inferred from Doppler shifting of the parent star (e.g., https://en.wikipedia.org/wiki/HD_209458_b).

As another example, the first evidence for dark matter came from observations in the 1930s of the Doppler shifts of galaxies in galaxy clusters, which suggested much more mass in the clusters than could be explained by the masses of the individual galaxies. Some of this "missing mass" was actually observed in the 1960s and 1970s, when orbiting X-ray telescopes showed X-ray emission from very hot, dilute gas within the clusters (unobservable from the ground because the Earth's atmosphere blocks X-rays). It turns out that the hot, X-ray-emitting gas has about five times the mass of the (stars in) the individual galaxies. So some of the missing mass has been found -- though you still need significant, as-yet-undetected extra mass in clusters to explain why they haven't flown apart long ago.

[throwawaymaths]

Good examples.