I'm not saying the tories won't win (I'm inclined to think they will) but that graph seems to be the sort of thing, like, for instance, no president being re elected with X% unemployment, that's a rule until it isn't.
It's quite possibly a spurious relationship, but there's a difference between noting a correlation and asserting a binary "rule" that's cherrypicked and questionable. Nate Silver looked into claims like that
for unemployment and found there was little evidence of a strong correlation between unemployment and Presidential votes, but particularly there was no reason to believe that there was some kind of "magic threshold".
