A bit of both. https://en.wikipedia.org/wiki/Gravitational_wave#Wave_amplitudes_from_the_Earth.E2.80.93Sun_system https://en.wikipedia.org/wiki/Gravitational_w

A bit of both. https://en.wikipedia.org/wiki/Gravitational_wave#Wave_amplitudes_from_the_Earth.E2.80.93Sun_system https://en.wikipedia.org/wiki/Gravitational_wave#Wave_amplit... runs through the numbers for the gravitational waves the Earth produces by orbiting the sun; it claims that at 1 light-year the amplitude would be 1 part in 1e26. For comparison, the wave that was detected is claimed to be "four one-thousandths of the diameter of a proton". That's about 7e-18 meters, on a baseline arm length of 4 km, so about one part in 6e20 -- about 175,000 times stronger than the waves Earth's orbit produces. And that was about 40x as strong as minimal sensitivity on LIGO, according to the article ("can detect changes in the length of one of those arms as small as one ten-thousandth the diameter of a proton"). Obviously if we were closer to the black hole collision we'd see much stronger waves. But you really do need very massive bodies accelerating very much (or equivalently orbiting very fast) to produce something that's detectable by LIGO over interstellar distances at all. The key part from this article is that the orbital period was about 1/250 of a second at the end; compare to Earth's orbital period. Going back to the formula given in the above Wikipedia entry, the frequency dependence is hiding in the "1/r" factor for the amplitude. 1/r is proportional to w^{2/3} (though it's not clear to me whether that's still true in a general-relativistic treatment; it's true enough for the Earth's orbit), which tells you how the wave amplitude scales with frequency...