Could be.
I've made my share of analytical models in my career. Models are used to make decisions, not necessarily to predict exact outcomes.
Models are intended to take uncertain variables and combine them in ways that differentiate decision-based alternatives under uncertain conditions.
The reason that models are less predictive is that low/base/high probability assumptions are used, not necessarily the entire continuous range of values. The best alternatives will stand out from the rest, even if the actual values are off from reality, because the same assumptions are used to assess all the alternatives.
This is what is meant by "directionally correct," meaning the alternative is correct even if the actual values are off.
A model alternative that suggests 2 million deaths will be worse than an alternative that suggests 100,000 deaths, even if the actual worst-case deaths might be 1 million deaths and the best case is 50,000 deaths.
That's why we don't focus on what "the model says," we focus on the preferred alternatives and the conditions that make them so.
-PJ