The model can be well correlated with reality based on data from after the event. They can be based on fundamental principles.
However if the values of the parameters are not based on the real/actual parameters of the current pandemic they can be all over the place on prediction.
Not enough testing to be able to nail the parameters of the current pandemic are going to provide inaccurate outputs of the models.
We do not have enough testing data to determine the parameters to put into the models.
GIGO.
We horrendously need to buy some more data.
The model was garbage.
When a model misses by a massive 25 times what they predicted only weeks ago, (like the Imperial College model that the crazy media have been using) that model needs to be thrown in the trash heap.