In February, as cases were falling rapidly, the disease modelers projected a large rise in daily new cases starting as soon as May (but with a multi-month window), and caused by the “UK B.1.1.7” variant.
The epicurve trended downward through July.
When it began to rise in late summer, the Delta or B.1.617.2 variant was prevalent, not the B.1.1.7 or “Alpha” variant as projected by the models.
Cases did rise at the very end of their modeled window – but not for the reason they gave (B.1.1.7) but instead for Delta. They modeled a completely different variant that ended up quickly fading.
They got a correct answer, almost, but for the wrong reasons. This does not make their model correct – in fact, the model was wrong.