Two relevant things:
First, Mike Brown and Konstantin Batygin comment on the Malhotra, Volk & Wang ArXiv e-print here:
http://web.gps.caltech.edu/~mbrown/papers/ps/findp9.pdf(Near the end of the paper.) Their final relevant sentence is this: “Thus, it appears that no useful constraint on the orbit or position can be drawn from this method.”
I’m not qualified to evaluate the arguments, but from their blog postings I’ve formed the impression that Brown & Batygin are very competent and cautious scientists and I’m biased in their favour.
The second thing is something that I discovered when I was fortunate enough to attend the British Astronomical Association one-day meeting on Robotic exploration of the Solar System on the 30th
April. (Incidentally James Canvin of UMSF gave a great talk on “Amateur use of Solar System spacecraft data”.) What attracted my notice Planet Nine-wise was a talk by Prof. Mark McCaughrean
on ESA’s Solar System exploration programme. (He is the Senior Science Advisor in the Directorate of Science & Robotic Exploration at ESA.) When talking about the Gaia astrometry mission,
he mentioned that the gravitational lensing effects of the sun and planets have to be removed as part of the data processing – the sun’s effects are practically 360 degree, and those for Jupiter are
very large, but I believe they do it for all the planets.
After the talk, I asked him if they could potentially detect the gravitational lensing effects of Planet Nine. He thought that they probably could – if they already knew where it was!
However, he thought that the volume of the data and the small signal would make it very difficult to extract without knowing where to look.
He was answering after just a few seconds thought, and I still wonder if there might be some hope in this method – after all, we should be able to calculate roughly what the lensing signal
would look like, and it’s moving, so Gaia, with multiple observations of the same patches of sky, should have observations with and without the planet, at multiple locations that have a
roughly predictable separation from each other. But I don’t know anything about the signal to noise ratio, or the data set. For a billion stars, I guess it’s pretty big!
Can anyone give a more informed opinion on this?