Yes, your right. and while I can't post any direct links I think I can post this.
Tan, G., Shaffer, F., Lyle, R., and Teo, I. (2016). Tan, G., Shaffer, F., Lyle, R., and Teo, I. (2016). Evidence-based practice in biofeedback and neurofeedback - 3rd edition. Wheat Ridge, CO: AAPB. Wheat Ridge, CO: AAPB.
You can find 1st edition online for free, but that is like 13 years old. The book lists each Condition, each study, each research backing up its claim. Granted it doesn't lvl 4-5 things all across the board, but it does as people have said seem to hold a lot of potential.
However lets say that people are right at the fact that "brain training" is snake oil. What about the EEG data itself? Its not like the computer is pulling it out of some hidden sub-file installed by programmers across the industry as a hidden joke to the rest of community. As a diagnostic tool EEG NFB in in my opinion still something I am fascinated with. Its quick, comparatively cheap (now, used to have to send the data off to a center and pay them to run a comparison, now the internet / computer's can do the job right at the office) and provides a measure of progress that isn't based on symptomatological increases or decreases but rather hard data that has minimal bias (unless people really can control their brainwave frequencies to such an extent as to alter the results when being tested which is what NFB claims to be teaching people in the first place).
I carry around my early EEG charts with me wherever I go. Whenever I endorse the treatment I always pull them out. Anyone can say "i tried this and felt better", but NFB was the first therapy where I could say "I improved, and here's the data to prove there was a change".