correctional measures applied
the neutrinos are generated from a tight beam of 2GeV protons, whose momentum constrains the neutrino beam, the neutrinos having very little transverse velocity. Its more complicated, of course, there are intermediate pions and a magnetic funnelling, if the link below describes what is actually done. (paywalled, abstract only) http://www.sciencedirect.com/science/article/pii/S0168900203007113
So, the focus is achieved by tuning the protons to have just enough energy to generate the neutrinos "at rest" in the moving frame of the protons (- i.e. near "c") - this eliminates transverse velocity and therefore produces a non-diverging beam.
Therefore there is a reasonable signal-to-noise in the detector - still only a few detections per bunch of 10^14 neutrinos, but enough for a picture to emerge. - that we can all argue about. The statistical analysis is hard, and may contain errors, hence the move to a "gapped" pulse, with a big feature in the middle, a more distinctive profile.