Should a recommendation engine ever tell you to do nothing?

I've spent the last few months building an FPL tool, and the design decision I keep going back and forth on has nothing to do with football.

Every recommendation product I've used is structurally incapable of saying "nothing here is worth doing". Streaming services always have something to watch. Trading apps always have a trade. It makes sense: the metric is engagement, and silence doesn't move it.

But in fantasy football the correct answer is genuinely "do nothing" most weeks, especially early in a season when you're looking at one or two games of data and mistaking noise for form. So I built the engine to say that. It scores every option, and if nothing clears a threshold over just leaving your team alone, the top card reads "Hold. Nothing here is worth your transfer" with the reasoning attached.

It felt wrong to ship. A tool that tells you to close it is a tool you might not reopen.

Then I ran it against real squads and it recommended holding for almost everyone, which is correct, and I realised it's the only version I'd still trust in October after it had been wrong a few times.

So, genuinely asking:

- Have you built something that recommends inaction, and did it survive contact with your metrics?

- Is there a version of this that isn't just leaving retention on the table?

- As a user of any recommendation product, would "nothing today" increase or destroy your trust?

Curious whether this is a principle or just an expensive opinion.