Thanks nad for the comments

One issue that I should clarify is that if some set of EOFs was created to model QBO, and these were based on some other phenomena or involved the QBO at different altitudes, it wouldn't really help much with an El Nino projection. Unless a fundamental equation is formulated or a simulation executed based on physical principles, there is no automatic way to extrapolate the fitted EOFs into the future. For example, if the EOFs are sinusoids no problem, but if the EOF is say, monthly rainfall in Wisconsin, it wouldn't help much. Perhaps that is being too pedantic on my part, but I have to occasionally remind myself of this argument to stay on the objective path.

You also mentioned the possibility of possible planetary effects. When I attempted a machine learning fit to QBO, one extended trial ended up like this
![qbofit](http://imageshack.com/a/img855/7435/femn.gif)

If you look at the frequency in one of the Fourier series compositions along the Pareto front, a frequency of 77.7 radians/year lines up with the lunar month synodic period of 29.5 days. And the 153 rad/yr is the half-month cycle. That could be just coincidental and why it is cool to get other people to cast skeptical eyes to the results.