Every restaurant ranking you've ever seen sorts on confidence, and confidence is mostly just volume. A 5.0 from nine diners is noise. A 4.6 from nine thousand is a fact. No algorithm can tell those apart without waiting for more people to turn up.
Which means the first page of results is, structurally, a report on where the island ate last year. That's fine for a safe dinner. It is useless for the meal you'll still be talking about in March.
The gap I book into
The places I'm interested in rate above 4.7, charge real money, and have almost no reviews. Perhaps forty of them at any one time. They're too new or too quiet to have accumulated volume, and they're the ones the chefs from the big rooms go to on their night off.
High rating plus high price plus low volume is not a red flag. It's a restaurant that hasn't been discovered yet.
Why you can't do this yourself
You can, roughly, but the failure rate is high, and on a nine-night trip you have maybe six dinners that matter. Getting two of them wrong is expensive in a way that isn't really about money.
What I'm doing is carrying the failure rate for you. I've eaten the bad ones. I know which of the quiet 4.9s is quiet because it's new and which is quiet because the owner's brother writes the reviews.
Planning something like this?
Tell me the dates and roughly how you like to travel, and I'll tell you honestly what I'd do with them.