I read it again, it clearly doesn't. ;)

As it turns out, we have multiple contradictory findings about whether high IQ is positively or negatively correlated with mental illness.

But even if you believe there's a negative correlation, it doesn't mean that many predictive SNPs can't predict both.

E.g. cloud cover is predictive both for rainfall and heavy snow. However, heavy snow is negatively correlated with rainfall. The latter doesn't negate the former.